Doing Corpus Linguistics

Citation

Csomay, E., & Crawford, W. J. (2024). Doing Corpus Linguistics (2nd ed.). Routledge.

Chapter Summary

The second edition (2024, about 180–190 pages) revises the 2016 first edition with fresh exercises, examples and references, an extensive list of English corpora around the world, new sections on identifying patterns in texts, and coverage of Cohen's statistical method (effect size). It has nine chapters in three parts and a closing list of available corpora. The authors write for a relatively novice audience: someone doing corpus linguistics perhaps for the first time, with limited or no knowledge of statistics. The book's theoretical frame is register analysis, which the authors credit to the work of Douglas Biber and Susan Conrad. To spare students from learning one search interface after another, the projects use the suite of online corpora created by Mark Davies, and the larger project uses one freeware program.

Part I, Introduction to Doing Corpus Linguistics and Register Analysis: Chapter 1: Linguistics, Corpus Linguistics, and Language Variation

Language as rules and systems, what corpus linguistics is (the analysis of authentic language), the difference between register, genre and style, and an outline of the book.

Chapter 2: Register (Functional) Approach to Language Analysis

Why register matters, what a register (functional) analysis is, how to describe situational characteristics and identify variables, how to give a functional interpretation, and units of analysis in register studies, with end-of-chapter exercises. Later chapters ask students to return to this framework when interpreting their findings.

Part II, Searches in Available Corpora: Chapter 3: Searching a Corpus

How to search for words, collocates, n-grams and part-of-speech tags in publicly available corpora, including corpora such as COCA and MICASE.

Chapter 4: Projects Using Publicly Available Corpora

A whole chapter of smaller projects, word- and phrase-based and then grammar-based, giving repeated practice with the same search interface.

Part III, Building Your Own Corpus, Analyzing Your Quantitative Results, and Making Sense of Data: Chapter 5: Building Your Own Corpus

Do-it-yourself corpora, choosing a corpus project (which takes considerable time), building the corpus with attention to copyright, and software programs for analysing it.

Chapter 6: Basic Statistics

Why statistical analysis is needed, basic terms, concepts and assumptions, and how to obtain statistical results, with exercises.

Chapter 7: Statistical Tests (ANOVAs, Chi-square, Pearson Correlation)

Difference tests and relationship tests, how to run them, and effect size, with exercises.

Chapter 8: Doing Corpus Linguistics

Carrying out a register analysis of one's own project in stages: situational analysis, linguistic analysis, functional interpretation and reporting. Worked examples include n-grams in problem-solution essays by Thai speakers of English. The publisher says the final project includes both a written paper and an oral presentation.

Chapter 9: A Way Forward

A short look beyond the book's methods to more advanced corpus research, including multidimensional analyses that identify co-occurring linguistic features through factor analysis.

Key Concepts

Part I (Chapters 1–2)

• Corpus: a principled collection of authentic texts analysed with computer tools.
• Register, genre and style: three ways of relating language to its situation of use, distinguished in Chapter 1.
• Situational characteristics: features of the context (participants, setting, purpose) that a register analysis describes first.
• Functional interpretation: explaining linguistic patterns by the situation in which they occur.

Part II (Chapters 3–4)

• Collocates: words that occur together more often than chance.
• N-grams: recurring sequences of words, such as four-word bundles.
• POS tags: part-of-speech labels that allow grammar-based searches.

Part III (Chapters 5–9)

• Specialized corpus: a corpus built to answer a specific research question.
• Normed counts: frequencies adjusted to a common text length so texts can be compared.
• Difference and relationship tests: ANOVA and chi-square for differences, Pearson correlation for relationships.
• Effect size: how large a difference is, not only whether it is statistically significant.
• Multidimensional analysis: identifying dimensions of variation from co-occurring features.

Critical Analysis

Strengths

• Written for beginners. The authors explicitly address first-time corpus users with little or no statistical background, and the statistics chapters start from why statistics are needed.
• One coherent framework. Register analysis runs through the whole book, from Chapter 2 to the project chapter, so students learn not only to count features but to interpret them functionally.
• Learning by projects. Chapter 4 is devoted to small projects in public corpora, and Part III takes students through building a specialized corpus and reporting a final project; using one family of online corpora keeps the technical overhead low.
• Updated edition. The publisher lists new exercises and examples, a list of English corpora around the world, new sections on finding patterns in texts, and effect size.

Limitations

• Short and introductory. At under 200 pages the book covers a few search types and basic tests; it does not teach programming, and Chapter 9 only points toward advanced methods such as multidimensional analysis.
• Tied to particular resources. The projects depend on Mark Davies's online corpora and a chosen software program; interfaces and access terms change, so some instructions may need updating.
• Mainly English and applied linguistics. The book is aimed at applied linguistics and TESOL students and its corpora list covers English; students of other languages or of computational approaches will need other sources.

Where It Fits in the Course

This is the gentlest introduction to corpus methods on the computational and corpus linguistics list. McEnery and Hardie's Corpus Linguistics: Method, Theory and Practice is the next step for the field's history, debates and range of methods. Gries's Quantitative Corpus Linguistics with R takes the statistics of Chapters 6–7 much further and teaches programming in R.

Language and Computers by Glass, Dickinson, Brew and Meurers is a non-technical introduction to language technology rather than corpus description. Bird, Klein and Loper's Natural Language Processing with Python and Jurafsky and Martin's Speech and Language Processing belong to the computational side of the course and suit students who want to move from searching corpora with ready-made tools to writing their own programs.

How to Use This Book

The book suits undergraduates and beginning postgraduates in applied linguistics, TESOL and English language, and language teachers who want to use corpus evidence. No statistics or programming is needed; familiarity with basic grammatical terms helps.

• Core: Chapters 1–4. Read Chapter 2 carefully, since the register framework is used in every project.
• For a research project: Chapters 5–8 in order. Start planning the corpus early, as the authors note that corpus projects take time.
• Statistics: Chapters 6–7 can be studied alongside Chapter 5 while the corpus is being built.
• Looking ahead: Chapter 9 and the list of available corpora.

Do every search yourself rather than only reading the examples, keep notes of search settings, and repeat the Chapter 4 projects with different words or registers. Plan time for the written report and the oral presentation of the final project.