📄 Johnston, Michael. 2023. The Middle English Book: Scribes and Readers, 1350–1500. Oxford: Oxford University Press. Pp. xvi + 288. ISBN 9780192871770.

Summary

This SELIM review by Cosima Gillhammer assesses Michael Johnston's monograph on how books were made and read in England between 1350 and 1500. The problem Johnston tackles is that scholars describe manuscript production with inherited labels such as professional, commercial, or amateur, and they use these labels inconsistently and on different kinds of evidence. Johnston studies a corpus of 202 manuscripts of four Middle English verse works, drawing on codicological, dialectal and archival evidence. His central idea is the closeness between the people who copied books and the people who used them. He proposes an intrinsic–extrinsic continuum and four descriptive book types. His main finding is that production in this period was decentralized and local. Bureaucratic institutions supplied scribes, and print later ended this closeness. The reviewer values the book because it replaces case studies with pattern-level evidence and offers a consistent taxonomy. She questions whether a verse-only corpus can support conclusions about other genres.

Key points

• The corpus comprises 202 manuscripts covering four works: Piers Plowman in every one of its versions, The Prick of Conscience, and Lydgate's "Dietary" and "Stans puer ad mensam." Texts copied alongside these works are also discussed.
• Chapter 1, "Nomenclature," introduces intrinsic vs. extrinsic production. The distinction asks whether a book was made inside or outside its first readers' circle, and it is treated as a continuum.
• The four book types are:
  • Elaborate Books: made in one campaign, likely extrinsic, heavy scribal infrastructure.
  • Streamlined Books: plain, needing little infrastructure.
  • Evolving Books: made discontinuously and loosely.
  • DIY Books: copied by the readers themselves.
• Chapter 6 uses Linguistic Profiles from LALME (Benskin et al. 2013). Most Prick of Conscience copies show provenance in the same region as their dialect, which suggests they stayed near where they were made.
• Chapter 7 argues that church, town and household bureaucracies recruited local scriveners, and these scriveners also copied literary texts. The book has 37 illustrations. The reviewer asks whether texts like the Wycliffite Bible or history writing would show the same patterns.

Interesting ideas

• I1. Johnston treats his core distinction as a graded scale rather than a binary, so a manuscript can be more or less intrinsic, which avoids forcing borderline items into the wrong bin.
• I2. The categories are framed as purely descriptive and cover most of the data, while outliers are openly acknowledged rather than squeezed in.
• I3. Two independent kinds of evidence, dialect and provenance, are compared, and their agreement becomes a finding in itself.
• I4. Inconsistent legacy terminology is replaced by categories defined through explicit criteria, here the proximity between producer and user.
• I5. The reviewer's critique is that conclusions drawn from one genre may not transfer, so corpus biases should be stated openly.

For Mova

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

• I1 (frames): Speech-act and irony labels in prag could carry a degree of strength instead of a yes/no tag, keeping uncertain cases visible and explainable.
• I2 (learn): Silver-data labeling could include an explicit "fits no class" outcome, so the trainer learns from clean cases and the outliers are logged for later repair rules.
• I3 (learn): Agreement between independent witness parsers can be treated as evidence for accepting a silver tree, and systematic disagreement as a signal worth investigating.
• I4 (tree): When converting between UD dialects or treebank conventions, relation labels could be mapped through explicit structural criteria rather than by name matching.
• I5 (answer): Reader evaluations on fables and school tests should report which genres they cover and add out-of-genre test sets before claiming general reading ability.