Reviewed by William O. Hendricks, Portland, OR

Widespread agreement exists that one critical component of a narrative text is the plot, which progresses from a beginning through a middle to an end. Michael Toolan’s study makes the reasonable assumption that there must be explicit features in the text that guide the reader in this progression. His hope, as laid out in his introductory chapter (1–13), is that corpus linguistic methods will make possible a semiautomatic identification of those features in a few representative twentieth century short stories. Ch. 2 (15–30) provides some background to T’s approach as well as a discussion of some methodological issues.

Chs. 3 (31–52) and 4 (53–76) focus primarily on James Joyce’s story “Two Gallants”. T uses the WordSmith Tools Keywords procedure to identify words that are disproportionately frequent, compared to a 500,000 word corpus of twentieth century fiction by major British and North American authors. The top keyword in the story is the name Corley. T isolates fifty-one sentences in the story containing this name; this abridgment makes evident, to a limited degree, the progression of the story. T concludes that such top keyword sentences are only one component of the textual features that signal progression.

In Chs. 5 (77–96) and 6 (97–112), T pursues a somewhat different approach to Alice Munro’s “The Love of a Good Woman”, which might more appropriately be termed a novella since it is 25,000 words long. One of several tactics T tries is dividing the text into twenty-six subsections and identifying the keywords that occur in each section. He then applies WordSmith’s Plotlinks to the keywords of each section to determine the number of occurrences of each keyword with other keywords. Sentences in which at least four different keywords cooccur are deemed especially relevant to story progression. Another tactic is to attend to the initial clauses in narrative paragraphs. T concludes that a focus on just these linguistic features is too narrow.

Chs. 7 (113–34) and 8 (135–64) present a multifactor model of narrative progression. Eight linguistic features, or parameters, are deemed to contribute to narrative progression. These include features already examined in the earlier chapters as well as (i) free indirect representation of a character’s thought, (ii) direct speech that has a transactional function, (iii) negation-carrying clauses, and (iv) mental process verbs that have a proposition as a complement.

Ch. 9 (165–88) deals with textual features that contribute to suspense and surprise. In the final chapter (189–200), T offers his own evaluation of his attempt to apply corpus analytic methods to the study of narrative progression. He recognizes that his corpus-based approach is still at a very early stage, and throughout this book he has been candid about shortcomings in his analyses. Nevertheless, T expresses optimism that future work will be fruitful since he sees no intrinsic defect in his method.

I remain unconvinced that corpus analytic methods represent the best approach for the analysis of narrative progression. T observes that these methods necessarily focus on textually explicit features rather than more implicit ones. However, plot structure can be fully uncovered only via inferences from what is explicitly stated. T occasionally makes reference to Gricean implicatures, but these references are insufficient. Only a developed system of textual inferences will make clear what explicit textual features need to be attended to.