Gender Biases and Linguistic Sexism in Political Communication: A Comparison of Press News About Men and Women Italian Ministers

This research on press communication uses a synchronic perspective concerning eighteen ministers, balanced by gender, in the Renzi government (in 2014), as well as a diachronic perspective concerning women ministers from five governments (from 2006-2014). The governments in 2014 and of 2013 were predominantly center-left, with the participation of center and center-right parties, whereas the previous governments had technical-professional rather than political ministers (in 2011), center-right (in 2008), and center-left (in 2006) ministers. In the synchronic analysis we explored the different ways in which the ministers are named, the relative presence of sexist/non-sexist, agentive/non-agentive, and abstract/concrete language in which they were presented. The first analysis comprised 332 headlines and the second comprised 1,356 headlines; we conducted a numerical and lexicographical analysis on the headlines. The results showed: more coverage for men than for women; gender biases in naming ministers involving a greater number of citations of women with both first and last name, whereas there were a greater number of citations of men with their first name only; the prevalence of sexist language that uses the generic masculine rather than the specific feminine (that is, the grammatical feminization of a typically masculine form) in representing women; an increment of the specific feminine in representing women in the last three governments over the previous two; no gender differences in the use of “I” and “We” as markers of agency; more quotations of direct discourse for women than for men; language slightly more abstract than concrete, for both men and women; more positive adjectives for women, and more negative adjectives for men. The results are discussed in relation to the international literature and to the Italian cultural-political context.

The political arena is typically considered androcentric.In fact, it not only shows a lack of women's presence, but also indicates the intervention of various dynamics and forces that continue to discourage the active participation of women in this domain.Among these dynamics and forces, an important role is played by the mass media, which often end up crystallizing the gap in favor of men, rather than trying to overcome it.It is not by chance that the IPU (Inter-Parliamentary Union, created in 1889, a permanent forum for political multilateral negotiations in the world) stated that the media was one of the main places through which to conduct a struggle for gender equality in the political space.In 1994, the IPU launched a "Plan of Action" that dictated the possible measures to promote actions in favor of women politicians i (Sensales, Areni, & Dal Secco, 2016a, 2016b).Nevertheless, with few exceptions, the media continued to favor a representation of politics as a place of masculine power.The different strategies through which they operate this role have been studied by the social sciences, thus allowing us to accumulate evidence of gender bias operating in political mediatized communication.These biases begin with the increased coverage of men politicians (Carroll & Schreiber, 1997;Kahn, 1992Kahn, , 1994;;Kahn & Goldenberg, 1991), and extend to the tendentious use of language that, through specific linguistic markers, results in obscuring the contribution of women (Reisigl, 2008;Sensales & Areni, 2016;Wodak, 2003).
In Italy, only a few studies have been conducted about media and gendered representations of politicians concerning both coverage (Sensales, Areni, & Dal Secco, 2016b) and linguistic gender biases (Basile, 2010;Pescia, 2010;Sensales, Areni, & Dal Secco, 2016b).From this last point of view the omission is particularly salient, considering that Italian is a grammatically gendered language, with masculine and feminine declination for the nouns, and other correlate forms, such as adjectives and pronouns with the same gender markers of the nouns to which they refer to.This renders the Italian language more sensitive to linguistic gender biases than the natural gender (e.g., English) and genderless (e.g., Chinese; see Prewitt-Freilino, Caswell, & Laakso, 2012) languages.With our study, we aim to address this lack of knowledge by focusing on political communication centered primarily on ministers of the Renzi government (politically oriented prevalently to the center-left, but also with the participation of center and center-right parties), and secondarily on the last five governments, leading up to the Matteo Renzi government.ities.This process is still in its first steps because of the persistence of gender discrimination, as with the example of the four women ministers who were without portfolio (this means that the ministers were at the head of less important ministries because they lack budget autonomy), whereas all of the men ministers were with portfolio.
The present survey study investigated how the Italian press answered this challenge that proposed openness towards a higher feminine presence in the political domain, which is traditionally androcentric.Towards this end, we used both a synchronic and a diachronic perspective.In the first case, we conducted a gender comparison of the representation of the 16 ministers of the Renzi government, while in the second case we compared the representations of the women ministers of the last five governments until that of Renzi, with the aim of examining possible evolution in the use of sexist/non-sexist language in relation to the different political orientations of the governments.Regarding the Renzi government, we tested the consistency of the recent trend, shown in part of the international literature, towards similar media coverage for the two genders.With reference to the linguistic dimensions, we explored the different ways in which the ministers are named, and the presence of a sexist/non-sexist, agentive/non-agentive, abstract/concrete language to talk about them.In this way, we will highlight the strategic use of language, which can make the role of women more or less salient in politics, both in a gender comparison referring only to the Renzi government, and in a temporal comparison, concerning the women only.

Biases in Coverage
International mainstream research has extensively studied the biases related to the different amounts of coverage provided to men and women politicians.In general, in the past, the results underlined a major coverage advantage for men over women (Carroll & Schreiber, 1997;Kahn, 1992Kahn, , 1994;;Kahn & Goldenberg, 1991), whereas recently, ambiguous tendencies have emerged.In fact, at the diachronic level, the coverage shows a trend towards a partial overcoming of the gap -favorable to the men -with more balancing (Bystrom, 2004;Bystrom, Robertson, & Banwart, 2001;Hayes & Lawless, 2015;Heldman, Carroll, & Olson, 2005;Jalalzai, 2006;Smith, 1997), and sometimes even covering the women more (Sensales et al., 2016b;Wiliarty, 2010).Such a trend, if confirmed in the present survey, could provide hope for the media to take an increasingly positive role in overcoming gender discrimination, as advocated by the IPU.This role is particularly important if we consider that experimental research provides evidence for significant disadvantages for women candidates who receive less media coverage than men (Kahn, 1992(Kahn, , 1994)).

Linguistic Gendered Biases
Alongside these studies, a developing critical approach is centered on the role of language as a powerful tool of sexist discrimination that is currently operating in mass media communications.This more innovative field, linked to the analysis of language, communication, gender, and politics (Wodak, 2003) has its foundations in the Austrian-German critical tradition of politics-linguistics (Politolinguistik, see Reisigl, 2008;Wodak & de Cillia, 2006).In Italy, this approach has stimulated a number of studies, many of them internal to the social representation tradition, which assume the opacity of language and the capability of masking power relations as revealed through the analysis of the context and of the specific linguistic devices that are bearers of gender discrimination.However, many Italian studies have focused primarily on women politicians (Sensales, Areni, & Dal Secco, 2012, 2013, 2016a), and only occasionally on a gender comparison (Basile, 2010;Pescia, 2010;Sensales, Areni, & Dal Secco, 2016b).

Naming Ministers
How the ministers are named in the media-namely, by surname or first name only, or both surname and first name -has been examined in studies showing at the international level that there is gender bias in this regard.
The women politicians are named with their first name only, whereas the men are more often named with their surname ii (Carlin & Winfrey, 2009).In Italy, some survey findings (Sabatini, 1987;Sensales, Areni, & Dal Secco, 2016b) illustrate the asymmetrical use of the first name, and of the first name and surname, for women and not for men in the press.Specifically, Sensales, Areni, and Dal Secco (2016b), in a study about the press representations of three pairs of men/women Presidents of the Houses of the Italian parliament in 1979, 1994, and 2013, found gender biases, such that the use of the full name (typically a way of naming with gender visibility) was higher for women than for men politicians.They also found that the over-utilization of the surname was more evident for men than for women.Then, in the legislature of 1994, the first name only, though rarely, was used for the woman President, while it was completely absent for the woman President in 1979, and for the two men Presidents in 1979 and 1994.On the contrary, in the 2013 government, there was a similar use of the first name for Presidents of both genders, interpreted as a general process of trivialization to make the politicians appear closer to the people.
Sexism in political mediatized communication has been analyzed in some Italian studies, revealing a general over-utilization of sexist linguistic devices.In particular, Italian, as a Romance language, is more sensitive than English to the use of these devices.The results of these studies point to the wide use of the generic masculine as an indication of the crystallization of a gap in favor of men engaged in politics, and the reduction of women's salience (Sensales et al., 2012(Sensales et al., , 2013)).In parallel, however, they have shown an increase over time of the nonsexist forms (specific feminine and epicene forms) (Sensales et al., 2016a(Sensales et al., , 2016b) as a possible signal of a tendency related to the major number of women parliamentarians.With the present research we try to examine if this increase is actually stable over time.

Agency
With regard to the concept of agency, Sensales, Areni, and Giuliano (2017) remind us that this was particularly studied during the 1970s and 1980s in the fields of linguistics and sociology, and later became a focus of social psychological analysis, cultural anthropological and media studies (Ahearn, 2001;Carli, 1990Carli, , 2006;;Hall & Donaghue, 2013;Okimoto & Brescoll, 2010).More specifically, Duranti (2004) pointed out that agency-the capability to control one's own behavior and the impact this can have on the behavior of others-can be manipulated through language, as in the case of specific grammatical markers such as the pronominal forms.In this case, the variations between impersonal forms and first person singular and plural indicate a variation in agency: from a very low to a higher level.Concerning the pronominal forms "I" and "we", in a study on political interviews, the pronoun "I" is considered a marker of subjectivity (see Zand-Moghadam & Bikineh, 2015), whereas experimental research has shown that it is mostly used by people with a lower social status, in contrast with the pronoun "we", which is most commonly used by people with a higher social status (see Kacewicz, Pennebaker, Davis, Jeon, & Graesser, 2014).
In Italy, Basile (2010) studied the role of agency in 18 interviews with men and women involved in politics, published  (Sensales et al., 2016;Sensales, Areni, & Giuliano, 2017, in press), differences in agentic linguistic styles were shown only in some cases.Pronominal use of the first-person plural was generally more frequent than the first-person singular.In this last case, gender differences appeared, with a greater utilization of this form by men than by women.This unexpected result was explained in light of the research of Bijeikienė and Utka (2006), where the "I" form was seen as a marker of personal responsibility, of the deictic center, of self-reference and confidence, and of personal involvement (Wilson, 1990), used by men MPs.Sensales, Areni, and Giuliano (2017) also regarded the pronoun "I" as a marker of subjectivity (see Zand-Moghadam & Bikineh, 2015), and for that reason it was evaluated as particularly agentive.From that perspective, Sensales, Areni, and Giuliano (2017) interpreted their results as related to the Italian context of a democracy in transition: increasingly centered on the leaders, specifically on an individual identity which would make a particularly salient focus on a subjectivity enacted by a masculine leadership.
Concerning the frequency of the usage of conditional verbal forms iii (a non-assertive modality, subject to the fulfillment of the given conditions), a signal of low agency, no gender differences were found between the different pairs of MPs under observation (Sensales & Areni, 2015;Sensales et al., 2016;Sensales, Areni, & Giuliano, 2017, in press).Sensales et al. (2013) assumed that agency could be studied through the direct voices of politicians quoted by journalists.They focused their attention on the Italian press, but only concerning women politicians.
Comparing the agency attributed by the journalists to the women ministers of the center-right, as opposed to the center-left, they showed that the reference of direct discourse privileged the women ministers of the center-right, thus attributing them with more agency.In other international surveys that compared the number of statements of men/women politicians reproduced in the news, the results showed that women were less likely to be quoted, remaining without a leading role (Aday & Devitt, 2001;Fernández García, 2015).As a result, men candidates were directly quoted in supporting their claims, giving the impression that they were more informed and qualified than women (Devitt, 2002).

Abstraction/Concreteness
With regard to the detection of linguistic abstraction, Semin and Fiedler's (1988) Linguistic Category Model (LCM) assumes specific inferential psychological processes that bind the issuer of the message to the recipient.Thus, the level of abstraction and concreteness of the terms used to describe people and events would vary depending on how, and where, the sender of the message intends to canalize the attention of the receiver.In other words, it depends on stable and durable characteristics of the agent (high abstraction level, such as for use of adjectives) or on contextual characteristics that are, therefore, external to the subject (high level of concreteness, as in the case of the use of descriptive action verbs).The model was applied to ingroup/outgroup confrontations.For the ingroup it was shown that positive evaluation was high in abstraction level, using adjectives that valorize the members of ingroup, whereas the negative evaluation was high in concreteness level, which attributes to the context the origin of negativity.A specular trend was found for the outgroup, with high concreteness levels for positive evaluations versus high abstraction levels for negative ones.
This result was replicated by Menegatti, Mariani, and Rubini (2012) and by Rubini and Menegatti (2014).In this research tradition there are few studies centered on linguistic abstraction as a means of discrimination against women.In particular, Menegatti et al. (2012) and Rubini and Menegatti (2014) examined negative linguistic judgments against women in the field of personnel selection.The results showed that the selectors were using linguistic abstraction as a tool to unfavorably represent women.In fact, they showed that compared to the men candidates, the women candidates were described using the most concrete positive terms (verbal forms) and with more abstract negative terms (adjectives).This favored cognitive inferences that made salient the context for positive evaluations, and the internal stable characteristics of the women candidates for negative evaluations.Particularly in the case of Sensales and Areni (2016), the data showed an attention to the situation rather than the person in 75% of cases.In this way, the women politicians were weakened and they were attributed little psychological salience.Concerning the valence with which the ministers were described, this emerged clearly when the journalists focused on the person.In these cases, the adjectives were negative only 8% of the time, whereas in 55% of the time the valence was positive.Thus the press showed an ambivalent attitude towards women ministers, on the one hand obscuring their role (focusing primarily on the situation), but on the other hand evaluating their personal characteristics more positively than negatively.Those results leave open the question of whether the trend obtained was gender biased, because there was no analysis of the representations of men ministers.The present survey attempts to address this lack of information by focusing comparatively on men and women ministers.

Objectives
Based on these different findings we analyzed political communication in the press with the following objectives, mainly of a comparative, descriptive type.For the ministers of the Renzi government we explored: (a) media coverage, detecting the number of news headlines dedicated to men/women ministers; (b) the frequency with which their first names and/or surnames were used in the headlines.This also contributes to gender visibility, which is either clearly manifested when both first names and surnames are used, obscured when only the surname is mentioned, or trivialized when only the first name is used; (c) the extent of sexist/nonsexist language used in the headlines for the eight women ministers.We considered generic masculine, non-concordance between article (with the definite article in feminine form) and nouns (in the masculine form), as well as dissymmetric feminine (with the feminine definite article before the surname) as sexist linguistic forms, whereas the neutral form (e.g., leader) and the specific feminine forms (e.g., ministra: woman minister) were considered non-sexist linguistic forms.Furthermore, we examined (d) agency, elicited by ministers in their direct discourse by using the first person singular or plural, or the corresponding verbal forms (individual subject "I" -less agentic -vs collective subject "We" -more agentic), and (e) agency, attributed by journalists to ministers by quoting, in headlines, their direct statements (" ").Finally, we examined (f) the level of linguistic abstraction detected by the LCM, applied to the headlines and entire articles, to verify the consistency between the linguistic categories used in the two fields.We studied the coherence of linguistic choices by editorial staff (for the headlines) and journalists (for the articles) to control, if we found incoherence, for the role played both by the cultural orientation of the newspapers and the gender of the journalist.Conversely, in the case of coherence, we accorded more stability to subsequent statistical analysis concerning the valence of adjectives (negative, neutral, or positive) aimed to explore eventual gender bias.
Additionally, for point (c), with the aim of studying the evolution of linguistic sexism, we propose a diachronic Renzi in 2014 (f 112).The last two governments were composed of a broad coalition, predominantly of the centerleft, but also with women ministers from the center-right wing.We mainly focused our analysis on the headlines for their capacity to act as "cognitive shortcuts" (Andrew, 2007) and "frames" (Condit et al., 2001;León, 1997) by guiding the reader's comprehension and by providing an interpretive framework for the news.In one case we analyzed contemporary headlines and articles to detect the linguistic variables of LCM.
Headlines were identified with the support of the electronic database Quid, from the Senate Library in Rome.In order to be included in the analysis, the ministers' surname and/or first name had to appear in the headline, and/or in the above-headline, sub-headline, and summary.

Procedures and Measures
For the data analysis, the headlines were imported into a textual file that is composed of all words in the headlines, plus the annotation of the direct discourse, and the modalities of content variable related to the LCM (detected both in headlines and in the entire article).The direct discourse was quoted with the acronym "DD" (Direct Discourse), whereas the modalities of LCM were quoted with the corresponding acronym (e.g. for adjectives in headlines "ADJ", and in the article "adj").In LCM and DD the acronym was followed by the initial of the surname of the minister in order to create a unique attribution for each minister.
For the use of sexist language, as well as the extent of its use, we detected the presence and frequency of the following three modalities in the vocabulary of headlines that referred only to women ministers: the generic masculine (e.g., Ministro, with the name in the masculine form); the non-concordance between name and definite article (e.g., la Ministro, with the feminine definite article before the masculine name); and the dissymmetric feminine (with the feminine definite article before the surname, e.g., la Boschi).In Italian, the latter is dissymmetrically used only for women and not for men.The use of non-sexist language, as well as the extent of its use, was identified by detecting the presence and frequency in the vocabulary of the headlines of the following two modalities: the specific feminine (e.g., Ministra, with the name in the feminine form); and the epicene form (e.g., titolare del Ministero, head [in Italian neutral form] of the Ministry, without any definite article).Concerning the specific feminine, a form that continues to be perceived as unusual and ungrammatical (Merkel et al., 2012;Sensales et al., 2016a), we considered cases where this form was used in a derisive, mordacious manner (Sensales et al., 2016a; see also the last two examples of headlines in section Results/Sexist/Non-Sexist Language).In these cases we interpreted the use of the specific feminine as a form of sexism.We thus controlled for the context of use of the linguistic form "ministra", with the aim of annotating its usage as sexist or non-sexist.For the diachronic comparison we referred to the presence of the generic masculine, the dissymmetric feminine, and the specific feminine, in journalistic communications about the last five governments until that of Renzi.In this case we did not control for the context of use of specific feminine forms, which will be analyzed in a future work.
We detected vocabulary in the headlines concerning the construct of agency.Particularly the pronouns "I" and "We", and all verbs in the first person plural and singular, were considered as signals of agency enacted by the ministers.Direct discourse (DD), in quotation marks, was viewed as a form of agency reported by the journalist (see Sensales, Areni, & Dal Secco, 2012, 2013).
For abstract versus concrete language, according to the LCM, we analyzed the different uses of language arranged in four increasing levels of abstraction.For this purpose a special analysis grid was applied to the entire article, and specifically to the headlines, to detect the modality of the content variable.We measured gender bias by comparing the news on sixteen men/women ministers.In the original model the level of abstraction started from the more concrete level, with descriptive action verbs (DAV, e.g. to walk, to talk, etc.; verbs referring "to a single specific action with a clear beginning and end, and with a physically invariant feature";  Coenen et al., 2006, p. 7).The latter were also analyzed according to their valence: positive, negative, or neutral.The grid for content analysis was applied to the headlines and articles by two independent, specially trained judges.In case of discrepancies in the rankings, a third judge reviewed the related encodings and made the final decision.
We then proceeded to create an index relating to the average level of linguistic abstraction.In this case we started with the indication in the LCM manual (Coenen et al., 2006, p. 15) that "each category can be weighted by a numerical value in order to obtain an abstraction score".Specifically, we assigned the number 1 for the DAV, 2 for the IAV and the SAV, 3 for the SV, and 4 for the ADJ.As stated in the model, to obtain the abstraction average score, the scores from each category should be added and divided by the total number of coded items in the description.Thus, this average degree of language abstraction varies between 1 (very concrete, only DAV's) and 4 (very abstract, only adjectives) and indicates how abstract or concrete a description is.(Coenen et al., 2006, p. 15).
In our case the average level was obtained by calculating the frequency distribution of modality of content variables related to LCM, detected for each minister in headlines and separately in articles.In the results section, we present some example of headlines for sexist/non-sexist language and for agency.

Newspapers Coverage
The coverage by gender confirms the presence of a gender bias in favor of men ministers.Men ministers were the focus of 57% of the headlines, compared with 30% for the women ministers.Finally, in 3% of the headlines, more women ministers received coverage at the same time, whereas in 10% of the headlines both men and women appeared at the same time in the headlines.

Naming Ministers
With the statistical package Spad-t for textual analysis (Lebart, Morineau, & Bécue, 1989), we used the Mots (Words) step that shows the frequency distribution of the overall vocabulary related to the headlines.With this step we obtained the frequency distribution of the names of each minister.Figure 1 presents the percentage trends in the use of first name, surname, and full name for women and men ministers.Generally, the ministers were named with surname only, in a percentage slightly higher for men (98.3%) than for women (94%).Whereas, the first name was only used for men in very few cases (1.72%), and was never for the women, underlining that the trivialized form was applied only to men.In contrast, 6.4% of the time only the women were cited with both first name and surname, in this way contributing to feminine visibility.Due to the distribution of marginal frequencies there were no conditions for the applicability of the chi-square: in fact, in 50% of the cells there was a theoretical frequency of less than five (see Agresti, 2013).

Sexist/Non-Sexist Language
With the Mots step of statistical package Spad-t, we detected the absolute frequencies of the use of sexist/nonsexist language in the headlines for the eight women ministers.Figure 2 provides the trends.The generic masculine was the linguistic form most often used (f 51), followed by the dissymmetrical feminine (f 28), and by the specific feminine (f 27, with the term ministra/e [woman/en minister] used 21 times).
We checked the context of how the word "ministra/e" (woman/en minister) was used to determine any negative connotations associated with it.In 11 cases it was used in a sarcastic or overtly negative way, whereas in two cases the context was neutral but associated with the aesthetic aspects and clothing of ministers, or with their maternal behaviors (breastfeeding).In contrast, in six cases the term was inserted into an agentive context.For the next categorization, related to sexist/non-sexist language, we re-attributed 13 cases to linguistic sexism and eight cases to linguistic non-sexism.The latter were added to the remaining six forms of the specific feminine.(2006-2008-2011-2013-2014) To test the hypothesis of a possible relationship between the uses of the three different linguistic forms in relation to the five governments we applied a chi-square test to the data shown in Table 1, and found a statistically significant difference (χ 2 (8, N = 960) = 78.3,p < .001).We then evaluated these results with Haberman's (1973) test for the significance of residuals.This test identifies the specific cells that contribute most to the chi-square test result (see Sharpe, 2015).The results indicated that, compared to 2006, use of the generic masculine increased substantially in 2011, use of the dissymmetric feminine declined in 2011, and use of the specific feminine increased in 2014.

Agency
With the Mots step we detected the frequencies of different linguistic markers related to agency.Taking into account the different coverage of men/women ministers to obtain the ratio of articles for each gender, thus allowing the comparability of results, we calculated the ratio between the frequency of the two types of agency (elicited by ministers using the first person singular and plural, and reported by the journalist quoting sentences of ministers) and the number of articles.
Figure 4 presents results that contradict some expectations based on the scientific literature: in the quotations of direct discourse, the ratio of agency for headlines related to men and women ministers indicated a higher level for women (.62) compared to men (.49)-and the differences were statistically significant (z = 2.17, p < .05).In contrast, use of first person singular and plural pronouns did not differ significantly by gender (z = 1.04, p > .05). and Federica Mogherini [Woman minister of Foreign Affairs] called the two soldiers: "Thanks for the dignity that you have shown."There is solidarity, but also affectionate words, from the women ministers."

Linguistic Category Model
The LCM showed more than 99% concordance between the two judges, with only twelve discordances on 2484 detections.Table 2 illustrates the frequency distribution of four modalities of the LCM for men and women ministers, both in headlines and separately in the entire articles.
We can make two different observations about these results: first, in general the abstraction level is high, with averages above 2.50 in three out of four cases; secondly, there was considerable coherence between the headlines and the articles concerning both women and men.In this latter case, although there was a slightly lower abstraction level for the men ministers in the headlines than in the entire articles, this difference was not significant.
Finally, regarding the frequencies of the adjectives' valence with all of the headlines and the articles in consideration, for both men and women there was a general prevalence of neutral adjectives.However, there were also more positive adjectives than negative ones (see Table 2).To test the hypothesis of a possible relationship between the positive/negative valence of the adjectives (without considering the neutral ones) in the headlines and articles and gender of ministers, we conducted a chi-square test on the data presented in Table 3.The results indicated greater use of positive adjectives for women ministers versus greater use of negative adjectives for men ministers (χ 2 (1, N = 253) = 6.79, p < .01).This result differs from expectations based on the scientific literature about LCM (Menegatti et al., 2012;Rubini & Menegatti, 2014).

Discussion
We focused on gender comparisons, which allows us to draw some conclusions about the role of media in promoting a vision of politics as a field that is becoming not merely masculine, but oriented toward a major symmetry between men and women present in the political arena.This role is not performed in a linear way, but through contradictory dynamics that sometimes see a prevailing androcentric perspective, and at other times shows a focus on the role of women.These contrasting elements may be the result of a changing reality that has not yet had time to be solidified.This is a reality in which in the last elections, there was a higher number of women in parliament than in the past.In this way Italy, with just over 30% of women parliamentarians (IPU, 2013), has jumped in rank, rising from 56th out of 140 countries in 2010 (Francescato & Mebane, 2011;Pacilli, Mucchi Faina, & Berti, 2012) to 29th place out of 188 countries in 2013 (IPU, 2013).
At a general level, in some cases the Italian press plays a role that reflects gender discrimination.One example of this is the practice of giving less space in the coverage of women compared to the coverage of men, confirming a trend already demonstrated by literature from other countries (Falk, 2009;Fernández García, 2016).This gender discrimination is also observable when the media promotes sexist language that favors the generic masculine over the other linguistic forms associated with women, when gender is made invisible because of the over-utilization of the surname, or in the naming of ministers.
In other cases, political news highlights processes that confirm a gender bias by simultaneously providing visibility for feminine identity, such as when using both the first and last name exclusively for women.Gender bias was also detected in relation to the men ministers, particularly with reference to the trivialization of language.Contrary to expectations, this process is not activated towards women, or in the same way towards men and women, as demonstrated in the literature (e.g., Carlin & Winfrey, 2009;Sensales et al., 2016b), but for men only.However, even for men politicians, the frequency of first name only references was pretty low.The reference to the first name highlights a dynamic functional to the logic of the mediatization/personalization of politics that, with this rhetorical device, can generate a sort of proximity between men politicians and ordinary people (Sensales et al., 2016b), confirming the popularization of an androcentric politics.On the other hand, the lack of proximity between women politicians and ordinary people can be interpreted as a signal that the women ministers are inserted by the reporters into an elite that can stimulate opposite reactions towards her distance from common people, as an elite that has a higher status requiring respect, or as a privileged group that nourishes forms of anti-politics.
Still, in other cases, journalistic communication reflects a subtle form of sexism using the specific feminine in a distorted manner, namely in a derisive or negative context.In this way a non-sexist linguistic marker is formally presented to mask a sexist approach towards women.Thus only 43% of the specific feminine forms have counteracted gender biases, whereas in the other instances gender bias was confirmed.This shows an ambiguity in the use of the feminine specific, with a prevalent negative function that illustrates well the opacity of language.
Thus, the media utilize the appropriate gender declination in an apparent sanitization (for the term see Ng, 2007) of the words that, nevertheless, supports the routinization of masculine dominance through a derisive, sarcastic context for women.
Concerning the results on linguistic diachronic comparison, the results show, for the last three Italian governments, an increase in the use of the specific feminine in relation to the previous two governments.The trend confirms the findings of previous work (see Sensales et al., 2016a) showing that this increase is parallel to the incremental presence of women in politics, and to the support of the Democrat Party (PD, the center-left party), a party that is particularly favorable to a balanced presence of women/men in the Parliament.In fact, in the last three governments, there was support of the PD: indirectly for the Monti government, and directly for the Letta and Renzi governments, in which the Heads of Government were from the PD.It is not by chance that only in the case of the Renzi government there was a statistically significant presence of the non-sexist linguistic form, underlining in this way the rooting of leftist culture in women's emancipation process (Sensales et al., 2013(Sensales et al., , 2016b)).This rootedness is also confirmed in the trends related to the center-left government of 2006, where the dissymmetric feminine had higher values than the values for this form in the other governments, in parallel with the lower values for the generic masculine compared to those in the other governments.The dissymmetric feminine contributes to the visibility of feminine gender also if it is a gendered form, because it is used for women only.Its usage shows a sensitivity of political communication by a leftist government towards the presence of women in politics.In our results concerning the Renzi government, gender biases were really counteracted when, in the headlines, the values referring to the use of quotation marks (direct discourse) by the reporter was higher for women than for men ministers.This is an unexpected finding that contrasts with the international literature.For example, Gidengil and Everitt (2000) note that in the television coverage of women Canadian leaders, the sentences of their speeches are quoted less frequently than those of men.The same authors arrived at similar conclusions some years later (Gidengil & Everitt, 2003).Also Aday and Devitt (2001), regarding press quotations concerning the USA Presidential candidate Elizabeth Dole and Republican men rivals, found that journalists reported sentences from her public speeches less frequently than those of the men candidates.The same results, related to the quotation of sentences and interpreted as signals of agency, emerge from other investigations conducted in Italy (Sensales et al., 2012(Sensales et al., , 2013)).On the contrary, in our study, it seems that newspapers have recorded and amplified the change taking place in the Italian parliament.This, in effect builds in part a counter-stereotypical representation of the women as more agentic than the men, whenever they quote more statements of women than men.How much this representation can stimulate overcoming gender stereotypes, or lead to more sexist attitudes, is an open question.In general, social psychological research has stressed that the deviation from role expectations is assessed negatively, especially when enacted by women (Eagly, 1987;Koenig, Eagly, Mitchell, & Ristikari, 2011).In this regard, a recent study in the political field (Sensales, Areni, & Chirumbolo, 2017), conducted in Italy, found an evaluation of lesser effectiveness of a hypothetical woman minister.This was characterized by counterstereotypical behavior (directive: agentic), compared to an evaluation of greater effectiveness when a woman minister engaged in stereotypical behavior (collaborative: communal).
Regarding the general use of pronominal forms, our findings showed no gender differences, contrasting with previous results of the analysis of interviews given by men and women politicians to the Italian press in 2009.In this analysis, Basile (2010) found gender differences, such that men were more agentive and more likely to use "we" than were women.In contrast, our results confirm other previous Italian surveys concerning the parliamentary context until 2009 (Formato, 2014;Sensales & Areni, 2015;Sensales, Areni, & Giuliano, 2017), where first personal pronominal forms occurred with similar frequencies in the speeches of both men and women MPs.The differences between the press analysis in 2009 (Basile, 2010) and the findings in Parliament pose a question about the relationship between the two contexts: the parliamentarian, more elitarian, in which gender stereotypes can be overcome in spite of the androcentric environment, and the mediatic, more popular one, in which gender stereotypes are rooted.In those two different realities, until 2009, the same politicians seemed to make specific linguistic choices, differently modulating the linguistic markers used in their speeches.In the sentences quoted in the headlines of 2014, these gender differences in linguistic behaviours disappeared.Both men and women politicians used first personal pronominal forms in the same way, a possible signal of a social evolution in which the same politicians felt openness towards feminine emancipation and acted linguistically in a consistent way, going beyond gender stereotypes.
Finally, comparing the level of abstractness/concreteness with which the ministers were presented, we noted coherence between headlines and articles denoting shared representations by redaction and journalists.In general, no gender differences emerged in the average level of abstraction.For most of the cases, the values were a little bit higher than those of theoretical averages.In contrast, there were statistically significant differences in the valence of the adjectives, with a greater positive valence for women than for men, who in turn were more frequently characterized with adjectives having a negative valence than women were.In this way, the gender expectations implied by the model of linguistic categories were contradicted.In fact, in their research, Menegatti et al. (2012) and Rubini and Menegatti (2014) showed that in personnel selection judges show considerable gender bias.They found that judges used linguistic abstraction (adjectives) to represent women with internally stable unfavorable characteristics, whereas for positive evaluations they used concrete terms leading to a focus on the situation.
However, it should be noted that a part of the scarce existing literature that compares men with women is concerned with fields other than the mass media and politics.On the other hand, the LCM is increasingly regarded as working in a socially determined situation that differently modulates the communicative purposes and, consequently, the level of abstraction and valence of the language (Douglas & Sutton, 2003;Semin, 2009).In our case, the newspapers showed a willingness to attribute psychological salience, with their focus on stable characteristics, to both men and women, with women represented by a decidedly more positive valence than their men colleagues.
Compared to a previous study (Sensales & Areni, 2016), conducted on women ministers in three different Italian governments (2006-2008-2011), the average level of abstraction has increased, as has the positive valence of the adjectives.However, it remains to be explored whether these positive assessments are linked to sexist qualities, such as a focus on aesthetic characteristics, or if they are related to stereotypical or counter-stereotypical characteristics.With this type of further analysis we will be able to disambiguate results that are not otherwise properly interpretable.At the moment we can only advance a possible explanation linked to the well-known "women are wonderful" effect (Eagly & Mladinic, 1994;Eagly, Mladinic, & Otto, 1991).This effect underlines that, in general, there are more positive attributions to women than men, especially when women are involved in stereotypical roles that, in our case, could be indirectly evoked from the references to their aesthetic aspect.

Directions for Future Research
An interesting direction for future research may be the study of possible presence of a Linguistic Intergroup Bias (Maass, Salvi, Arcuri, & Semin, 1989;Wigboldus & Douglas, 2007) in the body of articles written by men/women journalists to determine whether the information that favors ingroups over outgroups is expressed more abstractly (in our case men vs. women ministers), as suggested by the model.In this case, the detection of nouns may also be used as an indication of a higher level of abstraction, as proposed by some researchers (Carnaghi et al., 2008;Cichocka, Bilewicz, Jost, Marrouch, & Witkowska, 2016;Graf, Bilewicz, Finell, & Geschke, 2013).The frequency of noun usage should also be analyzed in relation to the political orientation of newspapers to verify if the rightist ones use them more than the leftist ones, as was found by Cichocka et al. (2016) in reference to grammatical preferences of conservative ideologies.
Future research may also examine whether the ideological-cultural orientation of newspapers influences the use of linguistic sexism.In particular, it can be assumed that the center-right daily newspapers use more sexist language and gender bias because of their roots in traditional gender stereotypical culture (Hershey & Sullivan, 1977), whereas non-sexist linguistic forms and a relative absence of gender biases are expected to be found more commonly in center-left dailies by reason of their higher sensibility towards feminist discourses, as has also emerged in other research (Sensales et al., 2013(Sensales et al., , 2016b)).Thus, with these future research directions, it will be possible to answer the many questions that this investigation has left open.
Previous governments were of the center-left in 2006 (Romano Prodi), of the center-right in 2008 (Silvio Berlusconi), a caretaker government (of technical-professional and not political persons) in 2011 (Mario Monti), and of the center-left with some parties of the center-right in 2013 (Enrico Letta).Concerning the Renzi government, it was, for the first time in the history of Italian democracy, balanced by gender (eight men, and eight women).This allowed us to study eventual gender asymmetries in mediatized political communication concerning the coverage, and the use of specific linguistic markers in a socio-cultural context in which politics tried to avoid some gender inequal- in 2009 in three daily newspapers.The findings show greater agency in the interviews with men (with over half of the men using the first-person plural), whereas just over half of the women used impersonal forms (which placed the agent in second order).Studying the Italian parliamentary debates,Sensales and Areni (2015) demonstrated different levels of agency in the parliamentary speeches made between 1976 and 2009 by two pairs of speakers with different political orientations.In the speeches of the radical pair (members of Partito Radicale [Radical Party]) -delivered mainly between 1976 and 1993-the man gave himself more agency than the woman, using the firstperson plural with higher frequency.In contrast, for the speeches of the leftist pair (member of Partito Democratico [Democratic Party])-delivered mainly between 1994 and 2009-it was the woman who more frequently used the first-person plural.In an analysis of four pairs of deputies' speeches during Italian parliamentary debates between 2001 and 2006 Sensales et al. (2012) andSensales and Areni (2016)  used the LCM to analyze representations of women ministers inProdi's (center-left, 2006), Berlusconi's (center-right, 2008), and Monti's (caretaker government, 2011) governments in the Italian press.The LCM was applied with the aim of identifying possible bias in the use of language related to different political orientations of the three governments.Results showed a general focus on situational factors, rather than on the stable characteristics of women ministers, independently of their political orientation.
comparison of the trends concerning the extent of generic masculine, dissymmetric feminine, and specific feminine in the news about women ministers in the last five governments up to 2014.The five governments have different politicalorientations-center-left (2006), center-right (2008), center (2011), or with a leading role of center-left and the participation of some center and center-right parties(2013 and 2014).For the analysis we use our data together with the published data from the survey ofSensales et al. (2016a), gathered according to the same criteria.Method MaterialsThe first analysis focused on 332 headlines, published in five Italian newspapers with different cultural-ideological orientations: L'Unità and La Repubblica on the left/center-left, Il Corriere della Sera of the center, and Il Giornale and Il Tempo on the right/center-right.The headlines included above-headline (brief introduction to the news), sub-headline (which clarifies and adds some detail to the news), and summary (which highlights some specific elements contained in the article).The time frame ranged from the first day the government was formed to 30 days later (22 February to 22 March 2014).We chose this temporal span because it corresponds to the initial phase of the government in general and of ministers in particular which, in theory, can highlight the expectations of the journalists more and any gender biases related to the novelty of the nominations (seeSensales, Areni, & Dal Secco, 2012).The diachronic analysis of the frequencies of sexist/non-sexist language concerning the women ministers of the last five governments comprised a total of 1356 headlines.For the four previous governments, we used data fromSensales, Areni, and Dal Secco (2016a), collected with the same criteria as our data (cf.Sensales et al., 2016a, p. 461).In this case, we had six women ministers of the center-left government run by Romano Prodi in 2006 (f iv 291), four women ministers of the center-right government run by Silvio Berlusconi in 2008 (f 159), three women ministers of the caretaker government run by Mario Monti in 2011 (f 480), and seven and eight women ministers respectively in the governments run by Enrico Letta in 2013 (f 314) and by Matteo

Figure 1 .
Figure 1.Trends in percentage of the use of first name, surname, and full names for women and men ministers.

Figure 3
Figure 3 presents the diachronic comparison of the usage of different forms of generic masculine, dissymmetric feminine, and specific feminine about women ministers in the five different governments from 2006 to 2014.The comparison indicates the prevalence of the generic masculine from 2008 to 2014, whereas in 2006 the sexist

Figure 4 .
Figure 4. Agency detected by pronouns used by ministers (First Pers.Sing.and Plur.) and direct discourse quoted by journalist in the headlines, in terms of the proportion by articles published for each gender.

Table 2
Frequency Distribution for the LCM Detected in Headlines and in Articles for Men and Women Ministers

Table 3
Distribution of Valence of Adjectives by Gender of Ministers