Metadiscourse is increasingly recognized as fundamental to the way in which writers construct text to achieve their communicative goals, but it remains a somewhat vague concept in certain respects, and the term is used in different ways by different scholars; so a full-scale treatment of the phenomenon is potentially of great interest. Ken Hyland has published a number of articles in recent years on metadiscourse, and he is clearly well placed to write this survey of the topic, designed, according to the blurb, as an “accessible introduction” for students of applied linguistics and teachers, as well as academics. The book is divided into three main parts: The first reviews previous attempts to define metadiscourse and proposes a revised definition and classification; the second, and longest, explores the functions of metadiscourse in a range of different types of text; and the third focuses specifically on the language classroom and suggests ways in which students can be encouraged to improve their command of metadiscourse.
The first part of the book is thought-provoking and presents a well-argued case for seeing metadiscourse as essentially realizing the interaction between writer and reader. One minor weakness is that Hyland is perhaps too concerned with highlighting what he sees as the flaws in previous models. Some of the problems for which he criticizes these models appear inherent to the phenomenon of metadiscourse (e.g., fuzziness, overlap between categories) and apply equally to his own model. However, he is fully justified in the emphasis he places on the fact that metadiscourse can be recognized only in context, and that the same linguistic form may be used in different ways, only some of which involve metadiscourse.
The central theoretical argument comes in chap. 3, when Hyland sets up the model of metadiscourse that is applied in the rest of the book, based on three key principles (outlined on p. 38). The first principle is that metadiscourse is distinct from propositional aspects of the text. This is very much in line with the accepted view, deriving from the first studies that introduced the term in the 1970s. Hyland rightly stresses that, as other analysts have found, this distinction appears neater in formulation than in practice, and it is not always a straightforward matter to decide exactly what counts as metadiscourse in a text (although in the second and third parts of the book the problems of identification – perhaps inevitably – tend to be downplayed). However, a model with explicit criteria for identifying the categories will help minimize the uncertainties, and Hyland's aim is to propose such a model.
The second principle, the key one, is that metadiscourse embodies writer–reader interactions. This leads Hyland to take issue with the traditional view adopted in studies such as Crismore & Farnsworth 1990, that the major division within metadiscourse is between interpersonal and textual categories. He uses the distinction introduced by Thompson & Thetela 1995 between “interactional” and “interactive” resources, and broadens the two terms so that the former covers most of the metadiscoursal features that other analysts place in the interpersonal category (these “involve the reader in the text”; 49), while the latter includes resources previously characterized as textual (these “help to guide the reader through the text”). This could be seen as mere relabeling, but Hyland's point is that all metadiscourse is interpersonal in nature, and that using these terms helps to highlight the functional complementarities of resources that earlier approaches tended to present as very different.
The third principle is that metadiscourse refers to “relations which are internal to the discourse” (38). This is based on the distinction (made by, e.g., Halliday & Hasan 1976) between external and internal conjunction: The former expresses “real-world” relations such as temporal sequence, while the latter signals the unfolding construction of arguments in the text. Expressions of internal conjunction are thus metadiscoursal. It is not clear whether Hyland sees this third principle as holding only for conjunctive relations or for all metadiscourse. The fact that in outlining the principle he also discusses the metadiscoursal function of expressions of subjective modality suggests that the broader view is intended. This is certainly the preferable option: If the somewhat ambiguous term “relations” is taken to refer not only to those between parts of the text but also to those between writer and reader, with “discourse” used in its wider meaning of the process of communication in which the text plays a role, the principle has a more far-reaching application and can cover metadiscoursal phenomena such as self-mention and engagement markers.
The model summarized on p. 49 is the core of the book. While it owes more to earlier studies of metadiscourse than is claimed, it is comprehensive and largely convincing. The categories are well grounded theoretically, and they form a coherent set: Hyland is particularly successful in suggesting why metadiscourse might draw on just these kinds of resources rather than any others. In addition, to a great extent the categories look workable, with clear recognition criteria for each. There is also a necessary warning that metadiscoursal signals are often multifunctional, realizing more than one category simultaneously.
However, I am not fully convinced by all details of the model. One point that I would query is the way in which the category of evidentials is fitted in. Hyland includes this under “interactive,” along with code glosses, frame markers, etc. However, the definition of evidentials on p. 51 does not indicate in what way they are similar to the resources in the other interactive categories – indeed, by describing them as “establish[ing] an authorial command of the subject” and “provid[ing] important support for arguments,” the definition points fairly clearly to their interactional character, since they are oriented toward the needs of the writer rather than of the reader. Evidentiality is generally accepted as being semantically close to epistemic modality, which, as Hyland argues, is clearly interactional in its function. Furthermore, the analyses in the second part suggest that evidentials work together with interactional resources: For example, in company annual reports, they are shown to contribute, with hedges and self-mention, to the ethos of the texts (“creating credibility appeals”; 74) rather than, as one might expect if they were interactive, to the logos (“creating rational appeals”). Apart from the fact that it would disturb the neat balance of having the same number of interactive and interactional categories, the case for seeing evidentials as interactive seems relatively weak. This is clearly an issue that will need further debate. (It is worth noting that reassigning evidentials to the interactional categories would mean that some of the claims in the second part of the volume about differences in particular genres based on the relative frequencies of interactional and interactive devices would have to be reconsidered.)
Whatever reservations one might have about this aspect of the model, the exemplificatory analyses in the main part of the book round out the picture in an extremely satisfying way and provide an excellent introduction to current areas of interest and methods of approaching texts. A range of research on metadiscourse in a variety of genres and languages is summarized and explained. Not surprisingly, the primary focus is on academic texts, from various forms of student writing to textbooks and research articles. This is the area in which Hyland has done most of his own research, and it is in the explorations of the functions of metadiscourse in such texts that the most rewarding discussion is found. Many of the ideas will be familiar to those who know Hyland's work; but it is very useful to have the studies brought together here, and to be reminded of the quality of the insights. In addition, a good deal of illuminating commentary arises in the process of synthesizing his previous work. For readers less well versed in the topic, this part of the book provides a lucid and appealing overview of the field, as well as models to follow in their own research. There is a satisfying balance of broad reviews of research with detailed accounts of particular studies. Other types of text are also brought in, including business reports, popular science texts, editorials, and Darwin's The origin of species. It would have been interesting to treat these more fully and to spread the net even more widely – for example, advertisements are only briefly mentioned, and there are many other genres in which metadiscourse would be worth exploring. However, the concentration on academic discourse is clearly justified in that one of the main motivations for research in this field has been application in the teaching of writing. The issue of application is expanded on in the final part of the book. The basic methodology put forward is the now standard one of raising students' awareness of rhetorical conventions; but Hyland comes up with some creative variations on this basic procedure.
Overall, the book thus fulfills its promise. From the research perspective, although I do not agree with all the positions adopted, the view of metadiscourse as embodying interaction is persuasive and represents a useful advance in our understanding of the phenomenon. The model proposed is coherent and robust, and the analyses provide empirical support for its validity and value. As an introduction to the topic, it is authoritative, clear, and engaging. The volume seems very likely to establish itself as the standard treatment of metadiscourse for the foreseeable future.
