Mapping the Unmappable: The Hidden Mathematics of James Joyce’s Finnegans Wake
How semantic mapping and fractal analysis reveal the astonishing mathematical architecture beneath literature’s most “unreadable” masterpiece
“riverrun, past Eve and Adam’s, from swerve of shore to bend of bay, brings us by a commodius vicus of recirculation back to Howth Castle and Environs.”
With these opening words — beginning mid-sentence and eventually connecting to the book’s final incomplete sentence — James Joyce established a circular narrative structure that has perplexed and fascinated readers for nearly a century.
Finnegans Wake (1939) represents the apotheosis of literary experimentation: a 628-page text that blends over 60 languages, creates thousands of neologisms, and constructs meaning through layers of puns, allusions, and portmanteau words. It’s been called incomprehensible, unreadable, and pointless.
But what if we could see it differently?
Using semantic mapping techniques — the same mathematics that powers modern AI language models — I built a system to visualize Joyce’s linguistic universe. What emerged was startling: Finnegans Wake isn’t chaos. It’s a mathematical structure almost indistinguishable from a pure fractal.
The Experiment: Mapping a Dream
To create a semantic map of Finnegans Wake, I employed a pipeline combining natural language processing and dimensionality reduction:
- Text Vectorization: 102 passages from all 17 chapters were transformed into 500-dimensional vectors using TF-IDF (Term Frequency-Inverse Document Frequency)
- Dimensionality Reduction: t-SNE projected these high-dimensional vectors into 2D space, preserving semantic relationships
- Clustering: K-means identified four distinct semantic clusters
- Mathematical Analysis: Calculated entropy, hazard functions, and multifractal spectra
The result? A map of Joyce’s mind.
Each point represents a passage. Colors indicate chapters. Notice how passages cluster by theme rather than by chapter — the circular narrative manifests in semantic space.
The Shocking Discovery: Literature as Fractal
In 2016, scientists at the Institute of Nuclear Physics in Kraków analyzed 113 works of world literature. Their finding about Finnegans Wake made headlines:
“The structure of Finnegans Wake was virtually indistinguishable from a purely mathematical multifractal.”
What does this mean?
A multifractal is a mathematical object exhibiting different scaling behaviors at different scales — like a coastline that looks similarly jagged whether you’re viewing it from space or examining it inch by inch. In text, this manifests as self-similar patterns in sentence length variability that repeat from individual clauses to entire chapters.
My analysis confirmed this finding. The multifractal spectrum of Finnegans Wake shows distinctive curvature indicating rich scaling behavior across multiple hierarchical levels.
This isn’t just a curiosity. It suggests that Joyce’s “stream of consciousness” technique — often dismissed as arbitrary — actually produces text with precise mathematical organization comparable to natural phenomena like turbulence, branching patterns in trees, and galaxy distributions.
The Decreasing Hazard Function: Sentences That Don’t Want to End
Here’s where it gets really weird.
The hazard function measures the probability of a sentence ending given that it has already reached a certain length. In normal text — including Shakespeare, Dickens, and Proust — the hazard function increases: the longer a sentence runs, the more likely it is to end soon.
Finnegans Wake is the only work of literature ever analyzed where the hazard function decreases.
Four views of Joyce’s mathematical architecture: sentence length distribution, hazard function, multifractal spectrum, and vocabulary richness by chapter.
This means the longer a Joycean sentence has been running, the less likely it is to end. His sentences gain momentum rather than approaching conclusion. They’re linguistic perpetual motion machines.
The technical measure is the Weibull shape parameter β. Normal text has β > 1. Finnegans Wake has β < 1 — and this property remains stable even when the text is translated into Dutch, French, German, Polish, or Russian.
Let that sink in: the mathematical signature survives translation. Joyce built something that operates at a level deeper than any single language.
By the Numbers: Joyce’s Linguistic Engineering
My analysis yielded these metrics:
Property Value What It Means Neologism Density 81.7% Over 4 in 5 unique words are Joyce’s inventions Hapax Ratio 78.1% 78% of words appear only once Word Entropy 8.51 bits Extremely high unpredictability Type-Token Ratio 0.476 Moderate vocabulary concentration Avg Word Length 4.45 chars Normal (the strangeness is semantic, not orthographic)
The neologism density of 81.7% is extraordinary. In a typical novel, you’d expect near zero. Joyce wasn’t writing in English — he was writing in a language he invented, using English as raw material.
The Portmanteau as Semantic Compression
Joyce’s signature technique was the portmanteau — a word blending multiple meanings, languages, and allusions. Consider this 100-letter monster from the opening:
bababadalgharaghtakamminarronnkonnbronntonnerronntuonnthunntrovarrhounawnskawntoohoohoordenenthurnuk
This “thunderword” combines thunder-sounds from multiple languages — Hindi, Japanese, Italian, Portuguese, Swedish, and more — representing the sound of the Fall (of man, of Humpty Dumpty, of Finnegan from his ladder).
Other examples from my corpus:
- commodius = commodious + Commodus (Roman emperor) + commode (chamber pot)
- passencore = French “pas encore” (not yet) + passenger + encore
- wielderfight = German “wieder” (again) + wield + fight
Each portmanteau operates as a compressed embedding — packing multiple semantic dimensions into a single token. This is exactly how modern large language models represent meaning. Joyce anticipated neural networks by 80 years.
The Semantic Heatmap: Finding Connections
Darker cells indicate chapters that are semantically similar. Notice how the opening (The Fall) connects to the closing (Ricorso), and how certain chapters form tight thematic clusters.
The heatmap reveals Joyce’s architectural planning:
- Opening chapters (Book I) establish vocabulary that recurs throughout
- Anna Livia Plurabelle (Chapter 8) occupies its own semantic territory — the lyrical river-chapter
- Book IV (Ricorso) connects semantically to multiple earlier sections, serving as recapitulation
- The circular structure manifests visually: end connects to beginning
Joyce as Engineer
Joyce famously described himself as “one of the greatest engineers” and characterized Finnegans Wake as “a wheel… a perfect square.” This sounds like artistic posturing. It wasn’t.
The mathematical properties we’ve identified — multifractal sentence structure, decreasing hazard functions, translation-invariant statistical signatures — indicate that Joyce created a text with mathematical properties approaching those of natural complex systems.
He intuited or deliberately constructed a linguistic system operating according to mathematical principles that wouldn’t be formally described for decades after his death.
Modern AI systems represent text using high-dimensional embeddings. Joyce’s portmanteau technique anticipates this: each coined word functions as a learned embedding, compressing multiple semantic dimensions into a single token.
Finnegans Wake might be understood as Joyce’s attempt to create a “universal embedding” — a text encoding all human experience in language pushed to its expressive limits.
Conclusion: The 300-Year Project
Joyce said he wrote Finnegans Wake “to keep the critics busy for 300 years.”
With computational linguistics, we’re beginning to understand the full scope of his achievement. The Wake isn’t merely a literary experiment — it’s a mathematical object, a self-similar structure rewarding analysis at every scale.
The semantic map reveals Finnegans Wake not as incomprehensible chaos but as a carefully engineered linguistic universe with its own topology. The mathematics — multifractality, decreasing hazard, translation invariance — suggests Joyce created a work transcending any single language, operating at the level of information itself.
“A way a lone a last a loved a long the”
— the final words, connecting back to “riverrun,” in an infinite loop of meaning.
What patterns would you find in your favorite “difficult” text? The mathematics of literature is just beginning to be explored.
About the Author: Brian Curry is the creator of PyCausalSim, Papilon and an AI, Data, and ML Scientist, analytics and Data Science professional with over 20 years of experience across enterprise companies, and founder of Vector1 AI, a part time side project developing tools and systems to move from correlation-based analytics to causal decision-making.
