computational stylistics for king lear

Lear Linguistic Model (LLM)

The Lear Linguistic Model (LLM) is a computational stylistics project that estimates how “mad” a passage from King Lear feels. It combines several linguistically motivated features into a single Madness Score between 0 and 100, then reports how that score is distributed across three stylistic dimensions.

Core linguistic dimensions

Semantic disorganization

Captures how much the passage jumps between ideas, images, or arguments. Higher values suggest weaker logical progression, abrupt topic shifts, or collisions of imagery that make the speaker sound mentally fragmented or unstable in their thinking.

Word-graph randomness

Treats the sequence of words and clauses as a simple graph of connections. Higher values indicate a more tangled flow of language – looping back, piling on clauses, and making sudden transitions – which feels chaotic rather than linear and controlled.

Lexical weirdness

Describes how unusual the diction is, especially in pronouns, vocatives, and punctuation. Higher values reflect intense questions or exclamations, strange address (“thou”, “slave”, “fiend”), and emotionally charged wording that pushes the speech toward theatrical or unstable territory.

Project contributors

The Lear Linguistic Model is a collaborative exploration at the intersection of literature, linguistics, and computation, designed specifically for close reading of King Lear.

Contributors:
  • Aldric Benalan
  • Lucas Horowitz-Kurtzberg
  • John Bridgeford
  • Bharat Bhati
  • Sagar Deshpande

The live analyzer uses these dimensions to estimate a Madness Score for any selected passage, and can visualize line- or chunk-level intensity as a heatmap across the scene.