English: The present study concentrates on the representation and the reception of gender stereotypes. The analysis was first carried out on an ad hoc corpus of cult romantic comedies and dramedies of AngloAmerican pop contemporary culture and secondly with a perception test. Both the corpus-driven analysis and the test results provide useful insights into the representation, recognition and entrenchment of gender stereotypes in language and in western culture. The preliminary findings generally confirm and validate the scientific literature, although showing some notable new elements.
Italiano: Il lavoro si incentra sulla
rappresentazione e la percezione degli
stereotipi di genere. La ricerca è stata prima
condotta su un corpus costruito ad hoc di
film cult della cultura pop contemporanea
anglo-americana appartenenti ai generi
romantic comedy e dramedy, ed in seguito
con un test di percezione. Il duplice
approccio utilizzato fa luce sulla
rappresentazione, il riconoscimento e il
radicamento degli stereotipi di genere nella
lingua e nella cultura occidentale. I risultati
si trovano in linea con la letteratura,
sebbene mostrino alcuni nuovi elementi.
In the era of digital revolution and screen
proliferation, movies have undoubtedly acquired,
thanks to their significance, a pivotal role in
shaping our worldviews. In fact, popular films have the
power to sway our collective imagination and
influence our attitudes on crucial issues related to
race, class, gender, etc. Characters in films reflect
and perpetuate the status and options of them in
today’s society and culture, and thus play an
active part in creating symbolic role models
Starting from the premises that gender is socially
constructed
The corpus. We compiled a corpus out of the
orthographic transcriptions of eight English and
American romantic comedies, using the web
software SketchEngine
Title Sliding Doors Billy Elliot Bridget Jones’ Diary Year 1998 2000 2001
Nation UK UK UK/USA
Bend It Like Beckham The Devil Wears Prada Juno Eat, Pray, Love Letters to Juliet 2006 2007 2010 2010 USA USA USA USA The resulting corpus is therefore a synchronic ad hoc corpus of 95,036 tokens. We further subdivided it into two subcorpora consisting of the turns of female and male characters – respectively 55,766 (58.7%) and 39,270 (41.3%) tokens (henceforth: M and F). We chose to gather a new corpus – instead of relying on existing ones – to obtain a higher control on the data. Moreover, popular romantic comedies are the perfect humus for a polarized representation of gender roles, because of their content and intrinsic structure. As will be seen, however, our results are comparable with the ones extracted from much the larger film corpus Cornell Movie-Dialogs Corpus.1
sis. We used the online text analysis software
WMatrix
Business: Selling As it can be seen, in our corpus women tend to speak about shopping, cleaning, personal care, and family, whereas men appear to discuss money, sports, work and male friendship. In table 2 are also present semantic domains which were relevant for both M and F speech, i.e. “Anatomy and Physiology” and “Intimacy and Sex” (in bold). These last two domains may emerge as strongly relevant due to corpus-specific reasons. Romantic comedies, in fact, are most often centred around romantic and quite physical relationships. However, what we think is of interest when analysing the overlapping between semantic domains between females and males is the different wording. Women and men refer to their bodies and their relationships in different ways, which are consistent with a polarization of gender roles (E.g.: breasts vs. boobs). Keywords are also worth mentioning. Their evaluation showed that women make larger use of politeness forms, while men resort to more swearwords and interjections, such as “right, all right”. 2 The stimuli-sentences were chosen to be as representative as possible of the entire corpus: they are evenly distributed among all the films of the corpus, with two or three instances from each film for each subcorpus.
Interestingly, the tendencies that emerged from
our small corpus are in line with Schofield and
Mehr (2016)’s analysis of the Cornell
Movie-Dialogs Corpus
The test. With the aim of testing the reception and entrenchment of gender stereotypes in speakers, we developed a perception test based on the results of our corpus-driven analysis. We manually extracted 18 lines per subcorpus2, each containing one or more of the stereotypical semantic domains and keywords that emerged from the previous WMatrix analysis. The resulting 36 extracted lines were used as stimuli in the perception test3. The choice of such limited number of sentences was determined by two reasons. The first, theoretically motivated, was not to repeat the same keywords and stereotypes too many times. Such repetition, in our opinion, could have influenced or biased the participants. The second reason, of a more practical nature, was to construct a reasonably-sized test to maintain participants’ attention and avoid fatigue, which could have influenced the responses. We extracted film lines containing a variable concentration of stereotypes, ranging from sentences referring to only one to several stereotypical domains. The selection was done manually, based on the rather obvious hypothesis that sentences more “stereotypically dense” would be recognised more easily. The stimuli-sentences were also chosen as deprived of context as possible, in order not to give any clue about the film of origin. Proper names were omitted, and when this was not possible, substituted with the string [XXX]. For example, in (1) the name of the male romantic partner was obscured so that the only clue to the gender of the speaker would be the linguistic stereotypes (shopping, mitigated swearwords, weaving).
1)
When [XXX] and I broke up for two weeks, I bought a loom, a frigging loom The test was presented to 22 native, bilingual or highly proficient speakers of English, 15 women and 7 men (mean age: 39.5). The task was to decide whether a given sentence had been uttered by 3 For reasons of space we do not include the complete list of the sentences extracted and used for the test. Several examples are reported in the text and in following footnotes. a man or a woman. In order not to force participants to a necessarily binary choice, the option “I don’t know” was also included. We additionally asked speakers to specify words, expressions or general concepts that influenced their answers. This provided us with interesting insights into participants’ process of thinking and categorizing.
Several interesting considerations arise from the analysis of the data. Firstly, it appears that overall the stereotypes were correctly spotted and categorized.
Chart. 1: Percentage of recognised stereotypes (in red) However, it also emerges that female stereotypes were more unambiguously recognisable, with fewer answers assigned to the other gender or to the “I don’t know” category (chart.1).
By examining more closely the results, a subdivision of the data can be made to account for the differences in it: recognised (more than 50% correct), ambiguous (between 25-50% correct) and completely misunderstood (less than 25% correct) stereotypes. Table 4 illustrates the distribution of answers in the three frequency slots.
> 50% 25-50% < 25% F LINES As was firstly hypothesized, sentences with a higher “density” of stereotypical keywords or semantic domains were usually the ones that speakers better recognised. Stimuli in the first group, therefore, consist of clear-cut and well 4 E.g.: “Give me the bag! I've got to get some proper shoes for the wedding now” (71%) (f); “What are you doing, eh? You're me best mate!” (82%) (m). 5 E.g.: “God! My mum had a fit when she saw the boots!” (47%) (f); “He's a kid. He's just a fucking little kid.” (47%) (m). 6 The reverse stereotypes utterances are the following. recognisable clusters of linguistic and conceptual stereotypes4. The second group is instead formed by stereotypes that were recognised by a substantial part of the informants, but not by the majority. This, in our opinion, may be due to several factors: some concepts, for example, could be perceived as less prototypical than others. In addition, some linguistic features (e.g. discourse markers) were not fully recognised as stereotypical due to our limitation to the written dimension. Prosody, contextual information and multimodality are in fact fundamental aspects of language that were inevitably excluded from our experimental design5. Finally, the last group consists of stereotypes that were not perceived as such by speakers (e.g.: family as a typical argument of women’s speech), and of what we called reverse stereotypes. That is, utterances that conceptually represented ambiguous events or anti-prototypical situations: a woman swearing, a man talking about his feelings.6 As predicted, these stereotypes were not recognised at all by participants, who tended to assign them to the opposite gender. It is interesting to note that also some male-produced sentences were not recognized by our informants, perhaps due to the composition of our corpus. Several predominant keywords and domains in M, in fact, may be strictly related to the chosen film genre. For example, the massive presence of the WMatrix domain Evaluation_inaccurate -- i.e. apologies --reflects the archetypical situation in romantic comedies of men apologizing for their mistakes to women. Being so context-related, however, speakers were not able to correctly locate sentences containing expressions from this domain.7 Another aspect that was taken into consideration in our analysis was the gender of the informants, to see if a relation with the data could be recognised. There was a statistically significant difference between the gender of the participant and the answer to the test (H (2) = 9.2388, p-value = 0.0024, Kruskal-Wallis test with Wilcoxon posthoc, Bonferroni p-value correction).
A chi-square test of independence was performed as well to examine the relation between gender of the speaker and responses given.
I. Oh, shit! I stubbed my foot on the side of the shagging bath! (f) II. This is the first time in 18 years I'm going to be able to call the shots in my own life! (m) 7- I made a mistake, such a big, BIG mistake and I'm sorry. I'm truly, truly sorry. - We accept that we fight a lot, and we hardly have sex anymore, but we don't wanna live without each other. The relation between these variables was significant. (χ2 = 10.298, p-value= 0.0058).
Chart. 2: mosaic plot of the results divided by gender. Chart 2 shows the difference in male and female informants’ answers. The numbers of the variable “responses” indicate the three possible answers of the test: “male” (1), “female” (2), “I don’t know” (3). As it can be seen, men assigned overall more utterances to the “I don’t know” option rather than to one of the two genders. Women, instead, show a fairly equal distribution of responses among the three conditions. Furthermore, both men and women assigned more utterances to female characters than to male ones (see table 5). This result is in line with the fact that women stereotypes were better recognised overall, in the sense that fewer answers were assigned to the other gender.
m 23% 30% f 29% 34% idk 48% 36% Table 5: distribution of informants’ answers divided by gender of the speaker Other useful insights into the data came from the words our informants identified as relevant to their decision. In fact, two tendencies emerged: speakers either indicated specific words, collocations or phrases, or answered with abstract concepts and pragmatic inferences based on the utterances. Interestingly, words and expressions exactly replicated keywords, while general and abstract concepts reflected the semantic domains that emerged in the corpus analysis. In addition, several speakers performed actual pragmatic inferences based on the stereotypical concepts contained in the sentences. For example, to (2) subjects reacted either with a specific word like in a) or with a more general consideration as in b). 2) Ooh, you must feel like you're about to find your long-lost soul mate! a) "soul mate" b) talking about feelings in general
The present paper proposes an original take on
investigating gender stereotypes in language. The
novelty in our approach lies in the hybrid
methodology that falls neither in the tradition of the
literature on “gendered discourse” nor in the more
recent field of corpus linguistics, but combines the
two and adds insights from psycholinguistics as
well. This kind of integrated analysis provided us
with preliminary results that help identify gender
archetypical roles, behaviours and linguistic
representations in modern western culture. What is
interesting to note is that the gender
representations coming to light from our corpus of
pop-culture films are based on features that are now
dismissed as clichéd and stereotypical by the
literature
The archetypical depiction of characters is particularly evident in popular comedies, which do not examine characters’ psychology in depth. The test validated our assumption that film language stereotypically portrays the way in which men and women talk drawing on recognisable traits attached to femininity and masculinity in our culture. In fact, speakers were mostly able to correctly assign the utterances to the right gender. In addition, all our informants showed metalinguistic –or second-level –awareness about stereotypical concepts and linguistic clues, and several of them also provided us with insightful and creative inferences based on the event described in the utterance. We interpret this as a sign of stereotypes being conceptual in nature, deeply entrenched in our representation of the world and accessed via linguistic clues. The “reverse stereotypes” also reinforce this idea.
Acknowledgements: We would like to thank professors Silvia Bruti, Alessandro Lenci, Belinda Crawford and Monika Bednarek for useful comments. We also gratefully acknowledge the reviewers for their valuable comments.