---
title: "Quantifying Prosodic Variability in Middle English Alliterative Poetry"
description: "The paper asks how to measure, as a number, how much the stress and alliteration patterns of Middle English alliterative verse vary within a…"
authors: ["Roger Bilisoly"]
year: 2015
venue: "arXiv:1501.03214"
source: "https://arxiv.org/abs/1501.03214"
doi: "10.48550/arxiv.1501.03214"
arxiv: "1501.03214"
licence: "arXiv — non-exclusive license to distribute"
licence_url: "http://arxiv.org/licenses/nonexclusive-distrib/1.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/bilisoly-2015-quantifying-prosodic-variability-middle-english/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/bilisoly-2015-quantifying-prosodic-variability-middle-english/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2015](https://md.research.elamur.ai/free/en/papers/2015/page/3/) / Quantifying Prosodic Variability in Middle English Alliterative Poetry

# Quantifying Prosodic Variability in Middle English Alliterative Poetry

**Authors:** Roger Bilisoly · **Year:** 2015 · **Venue:** arXiv:1501.03214 · **Licence:** [arXiv — non-exclusive license to distribute](http://arxiv.org/licenses/nonexclusive-distrib/1.0/)

**Links:** [Publisher's page](https://arxiv.org/abs/1501.03214) · [PDF](https://arxiv.org/pdf/1501.03214) · [DOI 10.48550/arxiv.1501.03214](https://doi.org/10.48550/arxiv.1501.03214) · [arXiv 1501.03214](https://arxiv.org/abs/1501.03214) · [OpenAlex](https://openalex.org/W1878709046)

**Citations:** 0 citations (OpenAlex, 2026-10-14)

## Summary
The paper asks how to measure, as a number, how much the stress and alliteration patterns of Middle English alliterative verse vary within a poem, and how to compare two poems on that score. Bilisoly codes poetic lines as strings in two ways. The first is a hand-made 0/1 vector that marks the alliterating words in each line. The second is J. P. Oakden's traditional notation, such as "aa/ax", together with his published line counts. Following Pennec's Fréchet-style generalization from Riemannian manifolds, he defines the "mean" as the observed string that minimizes the sum of squared edit distances to all other strings, and the "variance" as that minimum divided by the number of lines. The ratio of two such variances serves as an F-like statistic, and a permutation test supplies its p-value. On the first 220 lines of Piers Plowman and Sir Gawain, the binary coding shows no significant difference. Oakden's counts do show a significant difference. The method works with any distance, so it gives a general way to test variability for symbolic data.

## Key points
- Binary coding: 220 lines per poem. Piers Plowman's generalized variance comes to 1601/220, while Sir Gawain's is 1773/220, giving a ratio of 1.107. With 1000 shuffles of the 440 pooled lines, the two-tailed p is 0.094, so H0 (equal prosodic variability) is not rejected.
- Oakden coding: 2010 Sir Gawain lines and 7003 Piers Plowman lines across 16 meter types, after dropping the "complex groups" category. Both poems have "aa/ax" as their generalized mean. The variances are 35.33 and 193.15, so F_true = 5.47. None of the 1000 resampled F values reached it (the maximum was 5.07), so p ≈ 0 and the poems differ.
- The distance used is Levenshtein edit distance, which is a true metric. The mean is taken from the rows of the pairwise distance matrix and need not be unique.
- Alliteration is decided by sound rather than spelling. Simplified rules cover hard and soft c, g and yogh, pronounced initial clusters like kn-, and unstressed prefixes like "bi-".
- The resampled null distribution centers near 3.5 rather than 1. The author attributes this to the very unequal poem lengths and the heavily skewed meter frequencies.

## Interesting ideas
- For data with no arithmetic, you can still get a "mean" and a "variance": pick the observed item that minimizes summed squared distance to all others, and use that minimum as the spread; any metric can be plugged in.
- Two samples can be compared by the ratio of their spreads, with significance from shuffling the pooled items instead of a parametric F table.
- When the data arrive as a frequency table of types, the distance matrix between the distinct types can be weighted by their counts rather than expanding every token.
- The shuffled null came out centered far from 1, which shows that unequal sample sizes and skewed counts can bias a seemingly neutral ratio, so the null must be simulated, not assumed.
- The same phenomenon can be coded at two granularities: a per-word 0/1 vector and a compact stressed-word template. Running the test on both exposes how conclusions depend on the encoding.

## For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

- **I1** (tree): Using tree edit distance over the parses of one sentence from several parsers, we could pick the medoid parse and use the variance as a disagreement score.
- **I2** (learn): We could test whether a new parser or teacher tier produces more erratic output than the old one on the same corpus, with a permutation p-value instead of a gut feeling.
- **I3** (world): Variability of event orderings or frame patterns across many texts could be computed cheaply over distinct pattern types weighted by frequency.
- **I4** (learn): Evaluations with negative controls should always generate the null empirically under matched sample sizes, so that size effects are not mistaken for real gains.
- **I5** (back): A compressor could be checked at both levels, with token-level marks and abstract frame templates, to see whether round-trip variability is preserved at each level.

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

📚 Shelves (7): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 9,086 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 9,296 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 9,206 works · [🤝 Social Sciences](https://md.research.elamur.ai/free/en/shelves/social-sciences/) · 744 works · [🧮 Computational Theory and Mathematics](https://md.research.elamur.ai/free/en/shelves/computer-science/computational-theory-and-mathematics/) · 116 works · [🤝 General Social Sciences](https://md.research.elamur.ai/free/en/shelves/social-sciences/general-social-sciences/) · 87 works · [📚 Literature and Literary Theory](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/literature-and-literary-theory/) · 340 works

🧭 Topics (3): [English](https://md.research.elamur.ai/free/en/topics/english/) · 1,373 works · [🎙️ Speech & spoken language](https://md.research.elamur.ai/free/en/topics/speech/) · 233 works · [⏳ Historical linguistics & language change](https://md.research.elamur.ai/free/en/topics/language-change/) · 159 works

## Cite

Cite as: Roger Bilisoly. 2015. Quantifying Prosodic Variability in Middle English Alliterative Poetry. arXiv:1501.03214.

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@misc{bilisoly2015quantifying,
  title = {{Quantifying Prosodic Variability in Middle English Alliterative Poetry}},
  author = {Roger Bilisoly},
  year = {2015},
  doi = {10.48550/arxiv.1501.03214},
  eprint = {1501.03214},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.1501.03214}
}
```

- [← previous Perspectives on Arabic Linguistics XXIV–XXV, éd. Samira Farwaneh, Hamid Ouali, 2014, Amsterdam/ Philadelphia, John Benjamins Publishing Company](https://md.research.elamur.ai/free/en/papers/plonka-2026-perspectives-arabic-linguistics-xxivxxv-ed/)
- [next → Recipe, receipt and prescription in the history of English](https://md.research.elamur.ai/free/en/papers/bator-2019-recipe-receipt-prescription-history-english/)
- [↑ back to papers of 2015](https://md.research.elamur.ai/free/en/papers/2015/page/3/)

*Updated 24 October 2026*