University of Osaka
This paper describes a stylometric analysis of Agatha Christie's works with a special reference to the use of the feminine personal pronoun she in her novels. In a previous study, Tsuchimura (2016), as a result of the statistical analysis, it is shown that Christie tends to use the word she much more frequently than her contemporary Dorothy Sayers, a British mystery writer. She occurs about 12,000 times per million words in Christie's works whereas it occurs about 5,500 times in Sayers' works. A number of characteristic words were identified in the study, and this study focuses on the use of the word she in collocations within Christie's works.
It is possible to hypothesize that she occurs frequently in Christie's works because although the protagonists in Sayers' works are all male (Lord Peter Wimsey), Christie frequently employs female protagonists (Miss Marple and Tuppence Beresford). In order to test this hypothesis, a Random Forest (Breiman, 2001) classifier is trained on the 500 most common words from all of Sayers' works (55 texts) and Christie's works whose protagonists are not females (173 texts). As is shown in Figure 1, looking at the mean decrease in the GINI importance of the model per word, which measures the relative importance of each word in classifying a text as that of Sayers' or Christie's, the word she contributes strongly to classification of texts into 2 groups regardless of the protagonists' gender. We can see that Christie tends to use the word she frequently even in the works having or foregrounding male protagonists.
Figure 1. Variable Importance Plot of the result of Random Forests
The question arises as to how the word she is used in Christie's novels. To clarify its behaviour, this study examines collocations containing she in Christie's works. Following Sinclair and Jones (1974), who state that ‘for any node, a very high proportion of relevant information could be obtained by examining collocates at positions N-4 to N+4,' this study deals only with collocates at position N-4 to N+4. From a number of statistical measures commonly used to calculate the significance of collocations, this study chooses the MIscore, for ‘MI-score tends to give information about its lexical behaviour, but particularly about the more idiomatic (‘fixed') co-occurrences' (Hunston, 2002: 74). The minimum frequency of collocates was set at 10, and the threshold of the MI-score was set at 3.0.
The extracted collocates of she in Christie's works amount to 104, when proper nouns are excluded, the number of remaining types of collocates is 79. When compared with collocates of she in Sayers' works, those in Sayers' novels are mostly function words. It is thus difficult to see the behaviour of she in Sayers' works and in comparison with Christie's works.
When we compare collocates of she with those of he in Christie's works, we can see stereotypical femininity. The collocates of she in Christie works consist of a large number of words related to crying (sobbed, wept, wailed, screamed, crying) and actions indicative of fear or sadness (prostrated, frightens, fainted, shrank, clung, swayed, shivered). These collocates contrast with those of he, which include words such as hummed and laughs. Moreover, we can see from the collocates of she
and he that it is she who is choked or strangled by a male criminal while it is he who killed a female. This paper discusses in great detail how Christie describes female in her works from the perspective of collocates of she.
Bibliography
Breiman, L. (2001). Random forests. Machine Learning, 45: pp.5-23.
Hunston, S. (2002). Corpora in Applied Linguistics. Cambridge, UK: Cambridge University Press.
Sinclair, J. and Jones, S. (1974). English lexical collocations: A study in computational linguistics. In Teubert, W. and Krishnamurthy, R. (eds.) (2007). Corpus Linguistics: Critical Concepts in Linguistics. Vol.1(pp. 223-269). London: Routledge.
Tsuchimura, N. (2016, September). Stylistic Analysis of Agatha Christie's Works: Comparing with Dorothy Sayers. Paper presented at the sixth annual conference of Japanese Association for Digital Humanities, Tokyo, Japan.
If this content appears in violation of your intellectual property rights, or you see errors or omissions, please reach out to Scott B. Weingart to discuss removing or amending the materials.
Complete
Hosted at McGill University, Université de Montréal
Montréal, Canada
Aug. 8, 2017 - Aug. 11, 2017
438 works by 962 authors indexed
Conference website: https://dh2017.adho.org/
References: http://web.archive.org/web/20170802132745/https://www.conftool.pro/dh2017/sessions.php
Series: ADHO (12)
Organizers: ADHO