📄 Oliver, José María. 2026. A History of English. Perspectives and Case Studies. La Plata: Editorial de la Universidad Nacional de La Plata. Pp. 105. ISBN 9786316568816.

Summary

This is Sara Ennis's review of José María Oliver's short open-access textbook on the history of English. It is a humanities book review, not an NLP study, so it reports no data, models or metrics. Oliver built the book from his course materials at the Universidad Nacional de La Plata. Its aim is to show that languages always change and to explain how the varieties of English reached their present form.

The book has four chapters:

• the idea of linguistic change,
• change over time (periodization of Old to Modern English),
• change across space (Global Englishes),
• change in society (variation and change).

Each chapter ends with a hands-on case study built on authentic text samples. A recurring argument is that language is never neutral and is bound up with power.

Ennis praises the book's systematic synthesis of the literature and its discovery-oriented pedagogy. Her one reservation is that readers new to the specialist terminology may find it hard going. For a data or NLP team, the value lies in its framing of variation and periodization.

Key points

• The book has 105 pages, comes from Editorial de la UNLP (2026, ISBN 9786316568816) and can be downloaded free from the SEDICI repository.

• Chapter 1 surveys three schools of historical linguistics (neogrammarians, structuralism, generativism) and Schendl's (2001) three kinds of explanation for change (functional, psycholinguistic, sociolinguistic). Its case study, adapted from Campbell (2020), sets passages in Old, Middle, Early Modern and Modern English side by side.

• Chapter 2 compares the periodizations of Graddol (1997) and van Gelderen (2014). It argues that periods should be defined by combining external events with internal linguistic turning points.

• Chapter 3 reviews several models of World Englishes:

  • Kachru's (1992) three circles,
  • Jenkins (2015),
  • Kirkpatrick (2007), who classifies varieties by function: communication, identity and culture.

  The chapter also discusses standardization (Haugen 1996) and two negative consequences of English's global expansion: non-native speakers being pushed to the margins and a shrinking of linguistic diversity.

• Chapter 4 builds on Weinreich, Labov and Herzog (1968). It defines variable, variant, variety and style shifting, and stresses that variation does not necessarily lead to change.

Interesting ideas

• I1. Placing versions of comparable text from each historical stage side by side makes change visible and measurable, rather than just describing it.
• I2. Periods are best drawn by combining dated external events with internal structural milestones, instead of relying on either one alone.
• I3. Treating differences as variants of a single variable, not as errors, and remembering that variation need not mean change, keeps the analysis non-judgmental.
• I4. Kirkpatrick-style evaluation judges varieties by what they do (communicating, signalling identity, carrying culture) rather than by prestige.
• I5. Instead of supplying answers, the book poses anticipated questions and points the reader to primary source excerpts so they can work out the answers themselves.

For Mova

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

• I1 (tokens): A small stage-by-stage set of passages could serve as a stress test showing exactly where the tokenizer, tagger and lemmatizer break on archaic spellings and inflections.
• I2 (world): Event ordering in a text's world could link explicit dated happenings with inferred internal state transitions, giving a timeline that is easier to explain.
• I3 (tree): In UD dialect conversion and treebank merging, divergent annotations could be recorded as variants of one variable, so the parser does not learn that one convention is "wrong".
• I4 (learn): Evaluation sets could include texts from several Englishes and report results per variety, so a drop on non-standard input shows up instead of being hidden behind one "standard" score.
• I5 (answer): Reading-comprehension tasks could be framed the same way, requiring the system to cite the passage that supports its answer, which keeps every answer explainable.