Quantitative and Qualitive Analysis of Politeness and Gender Effects in Romantic Comedies
Citation:
Cabrera, Lucia, Quantitative and Qualitive Analysis of Politeness and Gender Effects in Romantic Comedies, Trinity College Dublin.School of Linguistic Speech & Comm Sci, 2022Download Item:
Abstract:
This study investigates differences in markers of politeness between male and female characters in romantic comedies. While focusing on the gender of the speaker, the impact of the gender of the addressee and of the screenwriting team on the estimated proportions of politeness signals was also explored. Measurable differences were observed for a number of indicators when considering the speaker role gender in isolation and in interaction with the gender of the listener and, in some cases, the gender of the screenwriting team.
The corpus analysed in this study contained scripts from 56 romantic comedies from 1977 to 2018. Script pre-processing techniques were created for each film in order to extract relevant information from each text. Regular expressions in R Studio and the politeness() package (Yeomans et al., 2018) were used to identify and extract politeness markers in the dialogue. Generalized linear mixed models were used to determine the effect of the gender of the characters, as speakers and listeners, and screenwriters in explaining potential differences in the use of indicators of politeness. These markers were analysed in context to determine their contribution to politeness in the dialogue. Overall, female characters had a tendency to use politeness markers more frequently than male characters, while the opposite is true for signals of impoliteness. In cases where the interaction between speaker and listener gender was statistically significant, characters displayed a higher estimated proportion of those markers when addressing the opposite gender.
A selection of scenes from three different films, written by different screenwriters, were analysed in detail in order to evaluate the extent to which the natural language processing tools captured politeness in text. A hypothesis of the features which most contributed to text being interpreted as polite within the NLP tool used was formulated. This process allowed for comparisons to be made regarding the classification of politeness and the identification of areas in which the computational tool worked efficiently and those in which it did not. The results showed that the tool used in this research was quite effective at capturing politeness, particularly in the cases where the most informative element was linguistic content. With regards to the classification of impolite text, the tool yielded inconsistent results. Impoliteness can be often realised by using elements beyond the text or which are highly sensitive to context, therefore these nuances are often lost when attempting to quantify them.
Differences observed between male and female characters align with more traditional views of language and gender. This research highlights the aspects where more significant contrasts were identified and how that affects the interpretation of text within a politeness framework. Using romantic comedies presents opportunities to investigate how contemporary issues are depicted in popular culture, particularly in a genre which attempts to mirror reality. The representation of men and women in film and the way they express themselves can have an important role in the construction of gender and therefore the exploration of these contrasts in inherently valuable. Finally, the generated corpus, script pre-processing techniques, and overall work pipeline represent a useful resource for subsequent research on this and other topics.
Author's Homepage:
https://tcdlocalportal.tcd.ie/pls/EnterApex/f?p=800:71:0::::P71_USERNAME:LCABRERADescription:
APPROVED
Author: Cabrera, Lucia
Advisor:
VOGEL, CARLPublisher:
Trinity College Dublin. School of Linguistic Speech & Comm Sci. C.L.C.S.Type of material:
ThesisCollections
Availability:
Full text availableKeywords:
politeness, computational linguistics, gender, film discourse, NLP, corpus linguisticsMetadata
Show full item recordLicences: