Daejeon, July 27–31
The Quan Tang Shi (Complete Tang Poems), compiled during the Qing dynasty, is the most comprehensive collection of Tang poetry, containing approximately 49,000 poems by over 2,000 poets. Despite this breadth, female-authored poetry remains numerically marginal, with only around 450 poems attributed to women. Traditional scholarship on Tang poetry has overwhelmingly focused on canonical male poets, while studies of female poets have tended to emphasize thematic interpretation and literary imagery rather than linguistic structure or stylistic patterning.
At the same time, Tang poetry includes a prominent genre, Guiyuan shi (boudoir lament poetry) and Gongyuan shi (palace lament poetry), in which male poets adopt a female speaking position to express political frustration, emotional restraint, and social marginalization. Existing scholarship on Guiyuan shi and Gongyuan shi largely treats the feminine voice as a literary convention or rhetorical device, rather than as a linguistic object of analysis. Consequently, the relationship between women’s self-expressed voices and male-constructed femininity has rarely been examined at the level of linguistic form.
More broadly, large-scale computational studies of Chinese poetry have tended to prioritize canonical male authors, while gender-based linguistic analysis remains underdeveloped. Therefore, this study seeks to demonstrate the value of corpus-based approaches for examining authorship, voice, and social identity in Tang poetry by establishing a carefully curated, small-scale corpus focused on gendered expression.
This study reframes femininity in Tang poetry as a phenomenon of linguistic construction. It asks how the feminine voice is produced at the levels of lexicon, syntax, and semantics, and how male poets’ constructions of femininity compare with women’s own linguistic practices. By foregrounding linguistic patterning rather than thematic content alone, the project shifts the discussion of gender from representation to structure. Drawing on historical sociolinguistics, this research treats gendered voice as a socially meaningful linguistic pattern shaped by power relations and literary conventions. Methodologically, it applies digital humanities techniques to a small but carefully curated corpus, emphasizing interpretability over statistical generalization. Substantively, the study examines the gap between women’s self-expression and male representations of femininity, revealing how gendered language in Tang poetry is conditioned by social position, authorship, and access to literary authority.
The corpus consists of two sub-collections: (1) poems authored by female poets included in the Quan Tang Shi, and (2) male-authored Guiyuan shi and Gongyuan shi that adopt a female speaking voice. Rather than pursuing large-scale statistical generalization, this project adopts a small-corpus approach suited to poetic texts, allowing for fine-grained linguistic analysis and contextual interpretation. The study combines computational analysis with close reading. Quantitative methods are used to identify recurrent linguistic patterns, while close reading situates these patterns within their literary and historical contexts. Specifically, the project employs: (1) Word frequency analysis to identify lexical patterns, including emotional vocabulary, kinship terms, and spatial expressions; (2) Part-of-speech (POS) tagging to examine syntactic tendencies, such as agency distribution and sentence structure; (3) Topic modeling (LDA) to explore thematic structures and their distribution across female-authored poems and male-authored Guiyuan shi and Gongyuan shi.
Close reading is conducted throughout to ensure that computational findings align with literary interpretation rather than replacing it.
The analysis centers on three linguistic dimensions: lexicon, syntax, and semantics. At the lexical level, the study investigates whether female-authored poems and male-adopted feminine voices differ in emotional expression, referential focus, and embodied language. At the syntactic level, part-of-speech distributions and patterns of agency and perspective are analyzed to compare how action, passivity, and subjectivity are constructed across the two groups. At the semantic level, the project explores how emotions such as longing, resentment, and resignation are conceptualized, and whether similar themes are framed differently depending on the author’s gender and speaking position.
Male-authored Guiyuan shi and Gongyuan shi thus serve as evidence of how femininity was imagined and stylized within male literary culture, while female-authored poems provide a baseline for examining women’s self-articulated voices. Comparing these two groups enables the study to identify both convergence and divergence in gendered linguistic practice in Tang poetry.
This research integrates gender studies, historical sociolinguistics, and classical Chinese literary studies through the construction of a specialized poetic corpus. By combining computational analysis with close reading, it offers a methodologically grounded approach to studying voice and social meaning in premodern texts. The broader goal of the project is to expand conversations about gendered representation, voice, and power in premodern poetic corpora by foregrounding women’s linguistic practices and critically examining male constructions of femininity. In sum, this study offers a new empirical perspective on gendered voice in Tang poetry through the analysis of a specialized corpus of female-authored poems and male-authored Guiyuan shi. By integrating digital humanities methods with traditional literary interpretation, the project demonstrates how computational analysis can enrich, rather than replace, close reading. Ultimately, this research contributes to a more linguistically grounded understanding of femininity in Tang poetry and provides a methodological model for gender-focused research in classical literature.