Mapping Literary Discourse: A Stylometric and Empirical Reflection on Corpus Stylistics and Machine Creativit

 

Mapping Literary Discourse: A Stylometric and Empirical Reflection on Corpus Stylistics and Machine Creativity


The intersection of computational linguistics and literary studies—broadly operationalized as Digital Humanities (DH)—has fundamentally expanded the methodologies of literary analysis. Rather than relying solely on traditional, subjective close-reading techniques, digital frameworks enable researchers to identify subtle linguistic patterns, character interactions, and narrative structures systematically across large textual datasets.

This post reflects on a series of empirical investigations using the CLiC (Corpus Linguistics in Context) web application, Voyant Tools, and an online experiment evaluating human vs. synthetic poetry. Through structured computational exercises, I evaluated speech clusters, contrasted fictional dialogue with authentic spoken corpora, analyzed Jane Austen’s narrative mechanics, and examined the stylistic boundaries between human intuition and machine generation.

Part I: The Human vs. Machine Poetry Experiment

To evaluate the perceptual boundaries between human-authored verse and synthetic text, I participated in an online poetry identification test comprising six anonymized poems. The task required categorizing each piece as either human-authored or algorithmically generated based on emotive depth, thematic coherence, structural cadence, and figurative language.

Diagnostic Results & Empirical Breakdown

  • Poem 1 (Human) — Correct

    Analysis: The poem displayed authentic affective resonance, featuring specific narrative anchors (e.g., hospital settings, maternal loss, and grief). The non-linear emotional cadence pointed strongly toward lived human experience.

  • Poem 2 (AI-Generated) — Incorrect

    Analysis: Misclassified as human-written. The model successfully deployed surreal imagery and an evocative atmosphere, demonstrating how generative systems can mimic imaginative depth through sophisticated semantic association.

  • Poem 3 (Human) — Correct

    Analysis: Correctly classified as human. The text exhibited a nuanced observation of nature—tracing the organic growth of a seedling—with a fluid, unforced rhythmic structure typical of intuitive composition.

  • Poem 4 (Human) — Correct

    Analysis: Correctly identified. The syntactic balance, measured pacing, and subtle handling of abstract themes (love and hope) reflected intentional artistic restraint.

  • Poem 5 (AI-Generated) — Correct

    Analysis: Correctly identified as machine-generated. While the poem adopted archaic poetic forms, the line-to-line thematic transitions felt mechanically forced, displaying structural formulaicness without organic semantic cohesion.

  • Poem 6 (Human) — Incorrect

    Analysis: Misclassified as machine-generated. The poem's extreme structural symmetry and rhythmic regularity mirrored the typical outputs of generative models, serving as a reminder that human poetry can also exhibit formal predictability.

Critical Reflection

Achieving an accuracy score of 4 out of 6 (66.7%) highlights the increasing sophistication of Large Language Models (LLMs) in replicating creative writing. Historically, literary critique assumed that affective nuance was uniquely human. However, modern computational models synthesize imagery and tone with striking fidelity. The core task in evaluating machine literature is no longer asking whether algorithms can write poetry, but analyzing how readers perceive, interpret, and attribute artistic intent to machine-generated discourse.

Part II: Corpus Stylistics via the CLiC Platform

Corpus stylistics bridges qualitative literary interpretation and quantitative corpus linguistics. Using the CLiC (Corpus Linguistics in Context) platform, our group examined narrative mechanics within Jane Austen’s oeuvre and the broader 19th Century Reference Corpus (19C), comparing fictional dialogue against real spoken English retrieved from the British National Corpus (BNC).

+-----------------------------------------------------------------------------------+
|                              LITERARY DATASET MATRIX                              |
+--------------------------+--------------------------------------------------------+
| Corpus Dataset           | Scope & Composition                                    |
+--------------------------+--------------------------------------------------------+
| 19C Reference Corpus     | 29 Nineteenth-Century English Novels                   |
| Austen Corpus            | Complete Novels of Jane Austen                         |
| BNC Spoken Corpus        | Authentic Spoken English (Transcribed Conversations)   |
+--------------------------+--------------------------------------------------------+

Activity 17.1: Speech Representation in the 19C Corpus

An analysis of recurring n-grams (speech clusters) across the 29 novels of the 19C corpus reveals how 19th-century authors (e.g., Dickens, Eliot, Hardy, Austen) constructed spoken discourse. While these dialogues convey conversational naturalness, corpus data demonstrates that fictional speech is highly stylized, functioning as a deliberate mechanism for characterization, social positioning, and plot advancement.

Pragmatic Alignment: 19th-Century Fictional Speech vs. Present-Day Spoken English

19th-Century Fictional ClusterModern Spoken EquivalentPrimary Pragmatic Function
I should like to knowI'd like to knowDeferential inquiry / Polite request
I should like to seeI'd like to seeSoftened desire / Implicit demand
I am sorry to saySorry to say...Mitigated negative assertion
What do you think of...?What do you think of...?Direct solicitation of opinion
It seems to me that...I think... / I feel...Hedged assertion / Stating a viewpoint
I am not going to...I'm not gonna...Direct refusal / Stating intent
What am I to do?What should I do?Expression of dilemma or distress

Corpus Concordance Insights

Concordance analyses show that 19th-century dialogue clusters prioritize grammatical completeness, social politeness, and formal syntactic framing. Unlike spontaneous spoken English, literary dialogue minimizes non-fluencies, using structured phrasing to negotiate class relations, etiquette, and interpersonal conflict.

Activity 17.2: Diachronic & Contextual Analysis of "it seems to me that"

Comparing concordance lines for the cluster "it seems to me that" across the 19C Novel Corpus and the BNC Spoken Corpus illustrates the distinction between literary construction and authentic spoken discourse.

                                  "IT SEEMS TO ME THAT"
                                           |
                +--------------------------+--------------------------+
                |                                                     |
        [19C Novel Corpus]                                    [BNC Spoken Corpus]
        - Formally Structured                                - Spontaneous & Fragmented
        - Grammatically Complete                             - Contains Hesitations (e.g., "erm")
        - Deliberate Characterization                        - Conversational Hesitation / Hedging

  • Shared Pragmatic Function: In both corpora, the phrase serves as an epistemic hedge—softening assertions, projecting politeness, and signaling subjective opinion rather than objective truth.

  • Structural Divergence: In the 19C corpus, the phrase occurs within complex, syntactically complete sentences. In the BNC Spoken Corpus, it is surrounded by conversational markers such as pauses, false starts, and hesitation tokens ("erm", "like").

  • Analytical Takeaway: Fictional speech operates as an idealized representation of dialogue. Authors capture the impression of natural speech while maintaining syntactic clarity and narrative drive.

Activity 17.3: Quantitative Analysis of Jane Austen’s Dialogue

An investigation into the top 15 speech clusters within the "Quotes" (dialogue) sub-corpus of Jane Austen’s novels illustrates her stylistic focus on decorum, interpersonal negotiation, and psychological interiority.

Top 15 Speech Clusters in Austen's Dialogue (Quotes Sub-corpus)

  1. I do not know what (Freq: 41)

  2. I am sure you will (Freq: 33)

  3. I do not know that (Freq: 25)

  4. very much obliged to you (Freq: 24)

  5. I have no doubt of (Freq: 21)

  6. and I am sure I (Freq: 20)

  7. I do not mean to (Freq: 19)

  8. I do not know how (Freq: 17)

  9. I do not pretend to (Freq: 15)

  10. I am sure I should (Freq: 14)

  11. I am sure I do (Freq: 14)

  12. quite out of the question (Freq: 13)

  13. and I do not know (Freq: 12)

  14. I do not think I (Freq: 9)

  15. I am sure I do not (Not found / Search variant)

Comparative Interpretation

While there is statistical overlap with the broader 19C corpus (particularly regarding expressions of cognitive uncertainty like "I do not know"), Austen’s dialogue displays a high concentration of epistemic markers ("I am sure") and polite hedges ("I do not mean to"). This pattern underlines the social dynamics of her novels, where characters continually navigate propriety, social obligations, and subtle disagreements.

Qualitative Concordance Validation

  • "I do not mean to": Functions predominantly as a face-saving pre-sequence, allowing characters to deliver criticism or express dissent without violating social decorum (e.g., "I do not mean to be impertinent").

  • "Quite out of the question": Functions as a decisive boundary-setting marker, signaling absolute refusal or social impossibility with minimal direct confrontation.

Activity 17.4: Character Speech & Epistemic Modality

Examining concordance data for specific clusters highlights how Austen constructs individual character voices and thematic tension through epistemic modality:

  • "I do not know what": Frequently deployed during moments of emotional ambiguity, social anxiety, or incomplete information, signaling a character's internal deliberation or tactical hesitation.

  • "I am sure": Operates as a double-edged stylistic device. While it overtly projects certainty and reassurance, concordance contexts often reveal underlying irony, self-persuasion, or social pressure.

Corpus stylistics allows researchers to scale up from individual close readings, providing macro-level empirical evidence for how recurring micro-linguistic choices construct narrative tone.

Activity 17.5: Dialogue vs. Narration (Quotes vs. Non-Quotes)

Contrasting the Quotes (character dialogue) and Non-quotes (narrative prose) sub-corpora in Austen’s works highlights a clear structural divide in her writing style.

+-----------------------------------------------------------------------------------+
|                        AUSTEN'S STYLISTIC DUALITY MATRIX                          |
+--------------------------+--------------------------------------------------------+
| Feature                  | Character Speech (Quotes)  | Narrative Voice (Non-Quotes)|
+--------------------------+----------------------------+---------------------------+
| Primary N-Grams          | "I do not know what"       | "as soon as they were"    |
|                          | "I am sure"                | "a quarter of an hour"    |
|                          | "I do not mean to"         | "in the course of the"    |
+--------------------------+----------------------------+---------------------------+
| Pronominal Focus         | 1st / 2nd Person (I, You)  | 3rd Person (She, They)    |
| Mode of Discourse        | Interactive, Evaluative    | Descriptive, Temporal     |
| Structural Function      | Social Negotiation         | Spatial & Plot Cadence    |
+--------------------------+----------------------------+---------------------------+

The Non-quotes cluster list is dominated by spatial, temporal, and relational phrases (e.g., "a quarter of an hour", "in the course of the", "the rest of the family"). These elements manage narrative pacing, track temporal progression, and establish social settings. Conversely, the Quotes sub-corpus focuses heavily on interpersonal stance, cognitive states, and social politeness, showing how Austen systematically alters her linguistic framework between character interaction and narrative movement.

Part III: Complementary Analytical Frameworks: Voyant Tools

Beyond CLiC, text analysis platforms like Voyant Tools offer complementary micro- and macro-analytic capabilities. By generating word-frequency distributions, trends graphs, and keyword collocation networks, Voyant allows researchers to visualize thematic shifts and lexical density across large collections of texts, complementing the concordance-based approach of corpus stylistics.

Pedagogical & Methodological Outcomes

  1. Methodological Integration: Combined traditional literary interpretation with empirical corpus linguistics, demonstrating how computational tools validate and expand close-reading hypotheses.

  2. Corpus Familiarity: Developed proficiency in leveraging the CLiC interface for concordance analysis, n-gram extraction, and sub-corpus comparison (Quotes vs. Non-Quotes).

  3. Sociolinguistic Awareness: Identified structural differences between fictional speech representations (19C) and authentic spoken discourse (BNC), framing literary dialogue as an artistic construction rather than a direct transcript.

  4. Stylistic Segmentation: Analyzed how Jane Austen systematically differentiates character dialogue from narrative exposition through specific syntactic choices and pronominal shifts.

  5. Critical Evaluation of Synthetic Text: Evaluated the capabilities and limitations of AI-generated poetry, developing a nuanced criteria for distinguishing human creative expression from computational pattern synthesis.

Works Cited

Comments

Post a Comment

Most Popular

Beyond the Colonial Cage: Globalization, Identity, and Capitalism in the Postcolonial World

  Globalization, Identity, and the New Postcolonial Condition Reading Dilip Barad’s “Globalization and the Future of Postcolonial Studies” and “Globalization and Fiction” through Ramin Bahrani’s The White Tiger Introduction: From Colonialism to Globalization Postcolonial studies traditionally examines the cultural, political, and psychological consequences of colonial rule. However, in the twenty-first century, colonialism can no longer be understood only through direct political occupation. Globalization has created new networks of economic power, cultural exchange, technology, migration, and political influence. The central question therefore becomes: Has colonialism really disappeared, or has it transformed into new forms of global power? Dilip Barad’s articles, “Globalization and the Future of Postcolonial Studies” and “Globalization and Fiction: Exploring Postcolonial Critique and Literary Representations,” provide an important framework for answering this question. Barad ar...
Read more

War Poetry: From Romantic Idealism to Modern Disillusionment

Image
What is War Poetry? Discuss its significance in the context of our classroom discussion regarding the content and form of war poetry. What is War Poetry? War poetry is a body of verse that takes war as its central subject , responding to its causes, experiences, consequences, and moral implications. It does not merely describe battles; rather, it explores human emotions, psychological trauma, patriotism, disillusionment, suffering, heroism, death, and the loss of innocence produced by war. While war poetry existed long before the twentieth century (for example, in Homer’s Iliad or medieval chivalric ballads), modern war poetry—particularly First World War poetry—marks a radical shift in both content and form . The unprecedented scale of mechanized warfare forced poets to abandon romantic ideals and adopt realistic, often shocking modes of representation . “The Soldier” – Rupert Brooke “Dulce et Decorum Est” – Wilfred Owen “The Hero” – Siegfried Sassoon “The Fear” – Wilfrid ...
Read more

God is Power: Political Religion and Absolute Authority in 1984

Image
  A Critical Exploration of Totalitarian Ideology, Faith, and Psychological Control in George Orwell’s Dystopian Vision Introduction  This academic task was assigned by Dilip Barad sir  , whose insightful teaching encourages students to critically engage with literary texts and multimedia resources. The assigned video, God is Power | 1984 | George Orwell , explores the central idea that in George Orwell’s dystopian novel Nineteen Eighty-Four , political authority replaces traditional religion and transforms itself into an object of worship. The video explains how the Party elevates Big Brother to a god-like status, demanding absolute loyalty, emotional devotion, and unquestioning belief from its citizens. It highlights the role of O’Brien as a symbolic priest of power who reshapes Winston’s thoughts through torture and psychological manipulation. The video further emphasizes Orwell’s warning that totalitarian regimes do not merely control people’s actions but seek to domi...
Read more