This book is a welcome addition to Cambridge University Press's excellent Studies in English Language series. For John Flowerdew it represents the culmination of work on a set of nouns that he first identified and named in 1994 and on which, often with PhD students and other research associates, he has been working ever since. The collaboration with Richard Forest has led to this, if not final, then at least most comprehensive account to date of those nouns.

The book consists of eleven chapters. The first five chapters set out the concept and theory of Signalling Nouns (SNs). The second half of the book is an extended study of a 600,000-word corpus of academic English and the SNs therein. The book also contains very extensive appendices, amounting to a third of the total length of the book, listing SNs and their relative frequencies, as well as details of the corpus. For reasons which I hope will become apparent I shall discuss chapters 6–11 before chapters 1–5.

The second half of the book, then, is an in-depth study of the SNs in a corpus of academic texts. The corpus is not large as corpora go (about 600,000 tokens) but it is carefully constructed. It comprises five disciplines from the natural sciences (Biology, Chemistry, Physics, Engineering and Ecology) and five from the social sciences (Economics, Politics, Law, Sociology and Business). The texts come from three genres: lectures, journals and textbooks. In each discipline/genre combination there are between two and six texts, with the number of texts selected to make each subcorpus around 20,000 tokens in size. This means that some of the subcorpora are very small indeed in terms of number of texts (two Business Studies lectures, for example). On the other hand, the comparability of the various permutations – comparing natural and social sciences, or journal articles with lectures and textbooks – is a distinct advantage.

Another advantage of the relatively small size of the corpus is that the authors have been able to identify and tag each instance of an SN. Corpus search and sort software is used to facilitate the procedure, but essentially the tagging is done manually. This is a very extensive task and numerous examples in the book illustrate the various distinctions that are drawn. The tags give the following information (summarized on pp. 73 and 74):

• • The semantic class of the SN, distinguishing between facts, ideas and locutions. Altogether six classes are identified.

• • Whether the SN operates anaphorically (referring back) or cataphorically (referring forward).

• • Whether the specification of the SN is to be found in the same clause or outside the clause. Where it is outside the clause, a further distinction is made between specification at a distance (‘distal’) and specification that is close by (‘proximal’).

• • Where the specification is within the clause, the lexico-grammatical patterning of the SN is identified. Examples include: complementation of the SN with a that-clause or to-infinitive clause; and clause patterns such as ‘SN + verb + specifics’. Some fine detail is given, resulting in about twenty-five different tags.

There is clearly a huge amount of detail here, and the tag system means that numerous comparisons can be made. Disciplines and genres can be compared in terms of the relative frequencies of the semantic classes of SN, the type of reference (anaphoric or cataphoric) and the lexico-grammatical patterns.

Following the description of method (chapter 6), there is a chapter giving an extensive set of examples of each of the categories (chapter 7), and then three chapters exploring the comparative frequency information: the distribution of the SNs (chapter 8), the distribution of semantic categories (chapter 9) and the distribution of the patterns (chapter 9). These chapters are full of interest, and if some of the information is less than surprising, it is only because previous work on SNs and similar concepts has led to the same conclusions. For example, SNs are used in the social sciences about twice as frequently as in the natural sciences. SNs in the ‘idea’ category are more frequent, proportionally, in the social sciences than in the natural sciences. Anaphoric relations are more frequent overall than cataphoric ones. Distal anaphoric relations are relatively infrequent in textbooks. The SNs equation, solution, method and problem are particularly frequent in the natural sciences, while right, policy, issue and theory are particularly frequent in the social sciences.

Careful attention to methodological detail, however, yields some unexpected results. For example, in comparing the overall frequency of SNs in the different genres, the authors point out that this frequency should be compared with that of nouns overall. Thus, whereas lectures have far fewer SNs than do journal articles, and about the same number as textbooks do, lectures also have fewer nouns overall, and in fact the proportion of SNs as against all nouns is higher in lectures than in either articles or textbooks (p. 101).

One of the difficulties in presenting work that is essentially about individual words in numerous contexts is moving between various levels of generalization. For example, one of the questions the authors ask is whether the natural-science/social-science divide operates consistently – do disciplines within each broad category use either the same SNs, or the same kind of SNs? To answer this question, they first classify SNs into six semantic classes (pp. 116–17). The classes are: ‘Act’ (process, change, attempt, reaction, etc.), ‘Circumstance’ (case, way, method, approach, etc.), ‘Fact’ (problem, thing, example, effect, etc.), ‘Idea’ (theory, idea, model, principle, etc.), ‘Locution’ (equation, question, point, argument, etc.) and ‘Modal fact’ (right, possibility, ability, need, etc.). Semantic groupings of this kind are always to an extent subjective and open to question. The distinction between ‘Idea’ and ‘Locution’ is particularly problematic, given that it is often unclear whether an argument or a prediction has been thought or said, and it could be argued that in academic discourse neither is very different from a possibility. Leaving that debate aside, however, there is then interesting information about the distribution of the categories across disciplines. In the natural sciences (p. 121), the proportions of the classes are relatively consistent, the exceptions being a higher than average frequency of ‘Circumstance’ in Biology and of ‘Locution’ in Engineering. The words in the ‘Circumstance’ group, or at least those that are frequent in Biology, belong to discourse about experiments (stage, method, technique, etc.), whereas those in the ‘Locution’ group, frequent in Engineering, are about calculations (equation, solution, formula, etc.). The same semantic classes are used to compare the social sciences, and Law here is shown to be the outlier discipline, with a higher proportion of ‘Locution’ and ‘Modal Fact’ SNs than the other disciplines, and a lower proportion of ‘Idea’ SNs. This is explained, not by the higher frequency in Law of nouns common to all social sciences, but by the specific vocabulary and phraseology of Law: caution, admission, legislation as examples of Locutions, and phrases such as right to silence and failure to give evidence as examples of Modal Facts.

The authors make a good job of identifying clear messages from the detail, and in distinguishing between disciplinary effects and genre effects. For example, they suggest that the higher overall frequency of SNs in the social sciences than in the natural sciences may be due to the preference in natural sciences for technical terms (p. 183). In other words, information tends to be ‘packaged’ in premodified noun phrases rather than in complex noun phrases or across clause boundaries. They note that some SNs are very frequent in some discipline/genre combinations: solution in Engineering textbooks, or strategy in Business textbooks. They suggest that the SNs that are distinctive of lectures (thing, point, way) are more the product of the mode of presentation than of the genre or the discipline.

The arguments made about syntactic pattern are similarly nuanced. Here the shorthand used to refer to the patterns requires careful attention, as does the argument. For example, CNAdj means ‘cataphoric SN in an adjunct, specifics in the main clause’; ANAcp means ‘anaphoric SN, across clause, proximal specifics’ (see p. 74 for a succinct but explicit explanatory table). In both these cases, the distribution of the pattern across subcorpora is similar to the distribution of SNs themselves, so the patterns do not distinguish disciplines or genres. A number of patterns are like this; others have a more marked distribution. For example, patterns including the ‘distal’ feature, indicating that the SN occurs at some distance from its specifics, are disproportionally frequent in all genres of Engineering and Physics, and in journal articles. An explanation offered for the latter is that the SNs occur in the introduction and conclusion sections of the papers.

There is, then, considerable value in the findings offered in this study of an academic corpus. Probably even more valuable, however, is its role as an exemplum. Other researchers, including PhD students, are likely to use the method as a template and compare other disciplines and genres with the results reported in this book. It would be interesting to extend the study to interdisciplinary academic domains, to draw comparisons with non-academic argumentative discourse, and to measure novice users of academic English against the lectures and textbooks that they mostly encounter as well as the journal articles to which they aspire (see, for example, Tåqvist Reference Tåqvist2016). The amount of detail given in the book, about both the tag system and the results, makes replication and comparative research very feasible and wholly rewarding.

Excellent though the second half of this book is, however, it may be the first part that proves the most useful to future researchers. In an expansive five chapters the authors discuss and exemplify the concept of Signalling Nouns in considerable detail. As well as giving examples of the most central nouns, about which there will be little disagreement, they explore the ‘outer edges’ of the concept, that is, the marginal cases, and in particular the nouns which are included within the category of SNs but excluded from similar conceptual categories. For example, they explain their rationale for including words such as paragraph or chapter (p. 61) and for excluding partitives such as kind of and type of (p. 62). They discuss the concept of abstractness and also explain their decision to include in the category of SNs those nouns which are abstract but not metalinguistic, such as activitism (p. 50). They clarify their exclusion of exophoric reference. They explain with great clarity the ‘bivalent’ SNs such as evidence, sign and way (p. 55). They explain in detail the semantic categories and the patterns which will underpin the research demonstrated in the second part of the book. In short, all questions along the lines of ‘I wonder how Flowerdew and Forest would treat this example’ are likely to find answers in the book.

This level of detail is of great value to researchers, whether they wish to find an exemplum to follow, or a point of comparison. For example, a researcher who wishes to explore nouns which require specification, but where the specification is to be found external to the current text (that is, cases of non-specific intertextuality), might use this book to explore the identification of such nouns, while specifying that these are not SNs in Flowerdew and Forest's sense. The clear benchmarking in the book is of equal use to researchers who wish to follow precisely in Flowerdew and Forest's footsteps and those who wish to set one or more of their criteria aside.

In general, the authors follow three precepts in defining SNs. Firstly, they make the category as broad as possible, including as potential SNs as many nouns as possible. Secondly, their identification of an SN is instantial; that is, a candidate SN may or may not be an instance of the category, depending on whether the noun requires and is accorded specification in a given context. This means that out of all the instances of a given noun in a corpus, only a proportion may be SNs. We are not given information as to what proportion of each candidate noun actually is an SN, but it is probable that some nouns (e.g. argument) are nearly always SNs, and those instances that are not may involve exophoric reference, whereas others, such as uproar or versatility, may be thought to be more likely to be rarely used as SNs. Finally, and linked to this, the identification of an SN is made on functional rather than formal grounds. Although a large of number of syntactic patterns are associated with SNs, it is the requirement for and presence of in-text specification that is the defining feature.

The authors specifically and generously acknowledge both the antecedents and the alternatives to their work. They are scrupulous in discussing the equivalences and differences between SNs and similar concepts, paying particular attention to Schmidt's concept of ‘shell nouns’ and Francis's notion of ‘anaphoric nouns’. Again, this careful comparison is very useful to researchers, whichever approach they choose to follow.

Is anything, then, missing from this valuable book? Inevitably, perhaps, the study of the academic corpus raises questions about the identification of SNs in other registers. It would be unreasonable to expect work of the same level of saturation on a larger or more diverse corpus, but a discussion of borderline examples in other contexts would be extremely useful. For example, how ‘abstract’ does an abstract noun need to be to make an instance of lexical cohesion also an instance of SN usage? I could have wished also for more extensive coverage of the literature that makes use of the concept of SNs, that is, literature that draws on Flowerdew's previous work, to complement the discussion of ‘competing’ concepts such as shell nouns.

I started reading this book somewhat sceptical of the notion of SNs. The work that I have done in the past has been with a set of nouns that is more narrowly focused both in terms of the structural patterns associated with the nouns and in terms of their meaning (i.e. ‘status nouns’), and it is difficult to abandon previous loyalties. Having read this book, I find myself more in sympathy than before with the very broad inclusion of nouns in the class, and with the functional criterion for identification, while acknowledging as the authors do that this makes searches of a large corpus difficult. I retain some doubts about nomenclature. SNs are not a class of noun (though similar concepts that are defined structurally, such as ‘shell nouns’, may be said to constitute a class). They are, rather, ‘nouns that (in these instances) signal’. This raises the more profound question of what it is that they signal. In some contexts, when the reference is across clause boundaries, the SNs organize discourse, so could be said to signal discourse structure. In those and other contexts, it could be said that the SNs label discourse: a stretch of discourse is an argument or a procedure or a point. They could then be said to signal discourse function or meaning – though ‘label’ seems to me a better word here. However, their defining feature is that they are specified – they are defined by what is done to them rather than by what they do (Specified Nouns). This is a minor issue for the ‘central and typical’ SNs, which both organize and label discourse, as well as being specified. It is a rather greater issue for the less typical instances of the category, which might be more likely to be found outside academic discourse.

In spite of this issue of nomenclature, I find that, thanks to this book, at the very least I now have a much clearer notion of what SNs are, and are not. It is a book I shall confidently recommend to any student who wishes to study SNs, or anything like them. The effect of the book has been more than that, though: I am now very much more persuaded of the value of the concept and its application.