Iberica 13 Ibérica 29 (2015): 15-34 ISSN: 1139-7241 / e-ISSN: 2340-2784 Abstract The concept of metadiscourse – the ways in which writers and speakers interact through their use of language with readers and listeners (also referred to as metalanguage and metapragmatics) – has received considerable attention in applied linguistics in recent years, particularly in the study of academic discourse. Conceptualised within the applied linguistics context of developing optimal descriptions of genres as a basis for a genre-based pedagogy, this article first reviews some of the different approaches to metadiscourse, highlighting how the concept is construed in different ways by different researchers. The article then discusses a number of problematic issues in metadiscourse research: metadiscourse as textual or interpersonal; the size of the linguistic unit in metadiscourse research; the multi-functionality of metadiscourse items; and the issue of representativeness in corpus research on metadiscourse. The second part of the article focuses on the concept of signalling nouns (SNs) (abstract nouns which carry particular pragmatic meanings in discourse), a feature of discourse not usually included under the rubric of metadiscourse. It is argued, however, that SNs represent an important resource for making writers’ (or speakers’) intended meanings clear. In this second part of the article, a first section introduces the notion of SN and a second section discusses how SNs might be incorporated into a model of metadiscourse. A final section of the paper concludes with a summary and some comments on pedagogic application. Keywords: metadiscourse, academic discourse, signalling nouns, corpus linguistics, genre, genre-based pedagogy. Revisiting metadiscourse: Conceptual and methodological issues concerning signalling nouns John Flowerdew City University of Hong Kong (China) enjohnf@cityu.edu.hk 15 03 IBERICA 29.qxp:Iberica 13 29/03/15 21:44 Página 15 Ibérica 29 (2015): 15-34 JohN FLowERdEw Resumen R e v is a nd o e l m ar c o de l me ta d is c u rs o : A s pe c tos c on c e p tua l e s y me to do l óg ic o s s o b re e l a ná l is i s de no mb re s re f e r e nc i al e s En los últimos años se ha prestado especial atención en lingüística aplicada al concepto de metadiscurso – el modo en que los escritores y los hablantes interactúan a través del lenguaje con sus lectores y oyentes (también referido como metalenguaje y metapragmática) – en concreto, al estudio del discurso académico. Este artículo revisa en primer lugar algunos de los distintos enfoques del metadiscurso conceptualizado dentro del marco de la lingüística aplicada para la descripción de géneros y como base para la pedagogía basada en género, destacando cómo este concepto está concebido de manera distinta por diferentes investigadores. El artículo aborda también algunos problemas relacionados con la investigación del metadiscurso: el metadiscurso textual o interpersonal; el tamaño de la unidad lingüística en la investigación del metadiscurso; la multi- funcionalidad de los elementos de metadiscurso y aspectos sobre la representatividad en investigación con corpus de metadiscurso. Esto concluye la primera parte del artículo. La segunda parte se centra en el concepto de nombres referenciales (NRs) (nombres abstractos que transmiten significados pragmáticos en el discurso), elementos discursivos que no siempre se incluyen dentro del marco del metadiscurso. Se argumenta, no obstante, que estos nombres representan un recurso discursivo importante para clarificar la información que transmiten los escritores (o hablantes). En esta segunda parte del artículo, una primera sección introduce la noción de NR y la segunda sección hace una valoración sobre cómo los NRs podrían incorporarse en un modelo de metadiscurso. La última sección del artículo concluye con un resumen y algunos comentarios sobre aplicación pedagógica. Palabras clave: metadiscurso, discurso académico, nombres referenciales, lingüística de corpus, genero, pedagogía basada en género. 1. Introduction In an earlier article (Flowerdew, 2011), I argued that a genre-based pedagogy in language teaching should focus not just on one type of meaning, meaning as action, which is the case in many situations, but also on two other types of meaning: content (propositional meaning) and identity (the way people use language to project their individual or professional persona). In this article, I would like to consider a further area of meaning, that of metadiscourse, which is also deserving of our attention if we are to come up with as complete descriptions as possible of discourse and genres that may 16 03 IBERICA 29.qxp:Iberica 13 29/03/15 21:44 Página 16 carry over to practice, especially in terms of identifying form-function relationships (while at the same time recognising that such relationships are variable and flexible). metadiscourse refers in very broad terms to “discourse about discourse”. The topic has been the focus of research for some forty years or more – at least since harris (1959). The field really developed, however, in the 1980s, with a number of important early theoretical contributions appearing during that period (williams, 1981; vande Kopple, 1985; Beauvais, 1989; Crismore, 1989). A further notable contribution was that of mauranen (1993). The field received a considerable boost to its popularity and a new methodological departure with two corpus-based monographs which appeared at about the same time in the 2000s: hyland (2005) and Ädel (2006). Since these two landmark publications, there have been further journal articles and a special journal edition devoted to the topic (Ädel & mauranen, 2013). At the same time, parallel work has been going on in metapragmatics (verschueren, 2004), which can be similarly defined as metadiscourse, although work on the former tends to focus more on spoken language, as opposed to written language, which tends to be the focus of work on metadiscourse. In addition, work in metapragmatics tends to be more theoretical, whereas work on metadiscourse often (at least more recently) adopts a more empirical approach. Finally, I may point out that metadiscourse falls under the more general rubric of metalanguage, the latter being one of the six functions of language identified by Jacobson (1960). The definition given above of metadiscourse as “discourse about discourse” is rather vague and indeed the whole notion of metadiscourse has been described as fuzzy (Ädel, 2006: 22-23; hyland, 2005: 16). All understandings of the term agree, however, that metadiscourse is concerned with non- propositional meaning. Communication can be viewed as occurring on two levels. There is the level of the propositional content and there is the metalevel of how to interpret the propositional content. metadiscourse refers to the latter level. The role of metadiscourse is thus to direct rather than to inform (williams, 1981). Its role is to signal the communicative intent of the writer/speaker and show how the propositional level of the discourse is to be interpreted on the part of the reader/listener. Some writers draw a parallel with halliday’s three metafunctions of language: ideational, interpersonal and textual. The ideational function corresponds to REvISITINg mETAdISCouRSE Ibérica 29 (2015): 15-34 17 03 IBERICA 29.qxp:Iberica 13 29/03/15 21:44 Página 17 propositional meaning, while the interpersonal and textual functions are the domain of metadiscourse. The interpersonal dimension of metadiscourse “helps writers express their personalities, their evaluations of and attitudes toward ideational material, shows what role in the communication situation they are choosing, and indicates how they hope readers will respond to the ideational material” (vande Kopple, 2002: 2-3). The textual dimension of metadiscourse is the dimension which organizes the text into a coherent message. while this distinction is helpful in highlighting two dimensions of meaning within metadiscourse, it nevertheless disguises the holistic nature of meaning and the fact that any stretch of text will carry with it simultaneously all three levels. This, indeed, is one of the problems when it comes to developing taxonomies of linguistic features which are associated with different aspects of metadiscourse, as we shall see. 2. Some models of metadiscourse If metadiscourse is a whole level of meaning, the counterpart to propositional meaning, then we will be dealing with a huge area of linguistic forms that have the potential to express, or realise, this level of meaning. various researchers have created taxonomies of metadiscoursal features, starting with vande Kopple (1985), who divided metadiscourse according to the two hallidayan metafunctions of textual and interpersonal, the textual category including text connectives, code glosses, illocution markers and narrators, and the interpersonal category being broken down into validity markers, attitude markers and commentaries. Table 1 shows these categories, along with glosses and examples of each. Crismore, markkanen and Steffenson (1993) revised this classification somewhat, keeping the textual/interpersonal classification, but further breaking down the textual component into textual and intertextual markers. hyland (2005) tried to break away from the textual/interpersonal classification, claiming that all metadiscourse is interpersonal insofar as it “takes account of the reader’s knowledge, textual experiences and processing needs and that it provides writers with an armoury of rhetorical appeals to achieve this” (page 41). The textual function, for hyland, is an enabling function which does not operate independently of the ideational and interpersonal functions (2005: 43). hyland (2005) classifies metadiscourse according to two major classes: interactive (to guide the reader through the JohN FLowERdEw Ibérica 29 (2015): 15-3418 03 IBERICA 29.qxp:Iberica 13 29/03/15 21:44 Página 18 text) and interactional (to involve the reader in the text). Both of these classes incorporate both interpersonal and textual functions, according to hyland (2005). Table 2 shows this model, including categories and examples. According to Ädel and mauranen (2010), two different traditions can be identified in the study of metadiscourse. The first of these, which includes REvISITINg mETAdISCouRSE Ibérica 29 (2015): 15-34 19 JOHN FLOWERDEW Ibérica 29 (2015): …-… Textual metadiscourse Text connectives to help show how parts of a text are connected to one another. Includes sequencers, (first, next, in the second place), reminders (as I mentioned in Chapter 2), and topicalizers, which focus attention on the topic of a text segment (with regard to, in connection with) Code glosses to help readers to grasp the writer’s intended meaning. Based on the writer’s assessment of the reader’s knowledge; the words used to define, explain, or delimit do not expand the propositional content of the text but help (no examples given except a gloss of a German word) Validity markers to express the writer’s commitment to the probability or truth of a statement, including hedges (perhaps, may, might), emphatics (clearly, undoubtedly), and attributors which enhance a position by claiming the support of a credible other (according to Einstein). Narrators to inform readers of the source of the information presented – who said or wrote something (according to Smith, the Prime Minister announced that). Interpersonal metadiscourse Illocution markers to make explicit the discourse act the writer is performing at certain points (to conclude, l hypothesize, to sum up, we predict). Attitude markers to express the writer’s attitudes to the propositional material he or she presents (unfortunately, interestingly, I wish that, how awful that). Commentaries to address readers directly, drawing them into an implicit dialogue by commenting on the reader’s probable mood or possible reaction to the text (you will certainly agree that, you might want to read the third chapter first). Table 1. Summary of Vande Kopple’s classification system for metadiscourse (adapted from Hyland, 2005:32). Crismore, Markkanen and Steffenson (1993) revised this classification somewhat, keeping the textual/interpersonal classification, but further breaking down the textual component into textual and intertextual markers. Hyland (2005) tried to break away from the textual/interpersonal classification, claiming that all metadiscourse is interpersonal insofar as it “takes account of the reader’s knowledge, textual experiences and processing needs and that it provides writers with an armoury of rhetorical appeals to achieve this” (page 41). The textual function, for Hyland, is an enabling function which does not operate independently of the ideational and interpersonal functions (2005: 43). Hyland (2005) classifies metadiscourse according to two major classes: interactive (to guide the reader through the text) and interactional (to involve the reader in the text). Both of these classes incorporate both interpersonal and textual functions, according to Hyland (2005). Table 2 shows this model, including categories and examples. REVISITING METADISCOURSE Ibérica 29 (2015): …-… Category Function Example Interactive Help to guide the reader through the text Resources Transitions Frame markers Endophoric markers Evidentials Code glosses Express relations between main clauses refer to discourse acts, sequences or stages refer to information in other parts of the text refer to information from other texts elaborate propositional meanings in addition; but; thus; and finally; to conclude; my purpose is noted above; see Fig in section 2; according to X; Z states; namely; e.g.; such as; in other words Interactional Involve the reader in the text Resources Hedges Boosters Attitude markers Self mentions Engagement markers Withhold commitment and open dialogue emphasize certainly or close dialogue express writer’s attitude to proposition explicit reference to author(s) explicitly build relationship with reader might; perhaps; possible; about; in fact; definitely; it is clear that; unfortunately; I agree; surprisingly; I; we; my; me; our; consider; note; you can see that Table 2. Hyland’s model of metadiscourse (based on Hyland, 2005) According to Ädel and Mauranen (2010), two different traditions can be identified in the study of metadiscourse. The first of these, which includes those models reviewed so far, takes a broader view and includes both the textual function (referring to the text itself) and the interactional function (indicating the relation of the writer to the reader); this model they label integrative, or interactive. The other model, they refer to as non-integrative or reflexive. This model is narrower and is restricted to what broader approaches classify under the textual function. In Hallidayan terms, the broad view includes both the textual and the interpersonal functions, while the narrow view includes the textual function only. (This view, it may be noted, seems to contradict Hyland’s argument that all metadiscourse is both interpersonal and textual). Representative of the narrow view and the focus on the writer’s commentary on the on-going text is Mauranen (1993) who prefers the term “reflexivity” to metadiscourse1. Reflexivity, for Mauranen, is divided into “highly explicit reflexivity” and “reflexivity of low explicitness”. Table 3 shows her categorisation, along with examples. 03 IBERICA 29.qxp:Iberica 13 29/03/15 21:44 Página 19 those models reviewed so far, takes a broader view and includes both the textual function (referring to the text itself) and the interactional function (indicating the relation of the writer to the reader); this model they label as integrative, or interactive. The other model, they refer to as non-integrative or reflexive. This model is narrower and is restricted to what broader approaches classify under the textual function. In hallidayan terms, the broad view includes both the textual and the interpersonal functions, while the narrow view includes the textual function only. (This view, it may be noted, seems to contradict hyland’s argument that all metadiscourse is both interpersonal and textual). Representative of the narrow view and the focus on the writer’s commentary on the on-going text is mauranen (1993) who prefers the term “reflexivity” to metadiscourse1. Reflexivity, for mauranen, is divided into “highly explicit reflexivity” and “reflexivity of low explicitness”. Table 3 shows her categorisation, along with examples. The final model I will refer to here is that of Ädel (2006). Ädel’s model, which, following mauranen (1993), she also refers to as reflexive, is a compromise between the broad and narrow views (Ädel, 2006: 180). Ädel defines metadiscourse as follows: metadiscourse is text about the evolving text, or the writer’s explicit commentary on her ongoing discourse. It displays an awareness of the current text or its language use per se and of the current writer and reader. (2006: 20) Following on from this, metadiscourse, for Ädel, can be divided into “metatext” and “writer-reader” interaction. metatext “spells out the writer’s JohN FLowERdEw Ibérica 29 (2015): 15-3420 JOHN FLOWERDEW Ibérica 29 (2015): …-… Reflexivity of high explicitness References to the text the paper, in this article, in the following section Discourse labels to illustrate, as noted earlier, this argument Addressing the reader note, recall, the reader Reflexivity of low explicitness Internal connectors however; second; also Discourse labels question; it is reasonable to think; (our present data) show References to the text now; as a first step Addressing the reader there is reason to remember Table 3: Mauranen’s “reflexive” model of metadiscourse The final model I will refer to here is that of Ädel (2006). Ädel’s model, which, following Mauranen (1993), she also refers to as reflexive, is a compromise between the broad and narrow views (Ädel, 2006: 180). Ädel defines metadiscourse as follows: Metadiscourse is text about the evolving text, or the writer’s explicit commentary on her ongoing discourse. It displays an awareness of the current text or its language use per se and of the current writer and reader. (2006: 20) Following on from this, metadiscourse, for Ädel, can be divided into “metatext” and “writer-reader” interaction. Metatext “spells out the writer’s (and/or the reader’s) discourse acts, or refers to aspects of the text itself, such as its organisation and wording, or the writing of it” (page 36). Examples given by Ädel from her data (page 37) are “I would like to discuss the topic”, “I stated above”, “I would like to emphasise” and “I will also give suggestions”. Writer- reader interaction “refers to linguistic expressions that are used to address readers directly, to engage them in a mock dialogue” (page 37). Examples given by Ädel from her learner data (page 37) are “but how did it start?”, “your second question may be”, “let me tell you the truth”, “well, why do I draw this conclusion?” and “[l]et’s first make a definition of the words”. Table 4, shows Ädel’s two basic types of metadiscourse further broken down into sub-types according to whether they are “personal” or “impersonal”, along with examples of each. 03 IBERICA 29.qxp:Iberica 13 29/03/15 21:44 Página 20 (and/or the reader’s) discourse acts, or refers to aspects of the text itself, such as its organisation and wording, or the writing of it” (page 36). Examples given by Ädel from her data (page 37) are “I would like to discuss the topic”, “I stated above”, “I would like to emphasise” and “I will also give suggestions”. writer-reader interaction “refers to linguistic expressions that are used to address readers directly, to engage them in a mock dialogue” (page 37). Examples given by Ädel from her learner data (page 37) are “but how did it start?”, “your second question may be”, “let me tell you the truth”, “well, why do I draw this conclusion?” and “[l]et’s first make a definition of the words”. Table 4, shows Ädel’s two basic types of metadiscourse further broken down into sub-types according to whether they are “personal” or “impersonal”, along with examples of each. 3. Some issues in metadiscourse theory In this section I will discuss a number of issues in metadiscourse theory. I have selected these issues for discussion, first, because they are important in themselves and, second, because I will return to them later in my discussion of SNs. REvISITINg mETAdISCouRSE Ibérica 29 (2015): 15-34 21 REVISITING METADISCOURSE Ibérica 29 (2015): …-… METATEXT WRITER-READER INTERACTION Personal Personal Impersonal Text/Code- oriented Participant- oriented Writer- oriented Reader- oriented Participant- oriented Reader- oriented T hirdly,… T he above-m entioned list… In other w ords,… T he question is… A definition of… involves… T his essay w ill prim arily deal w ith… as w e have seen… in our discussion above… if w e take… as an exam ple… A s I have show n… M y conclusion is that… B y… I m ean… A s a w riter, I w ould like to argue that… as you have seen… Y ou m ight w ant to read the last section first… I know you think that… C orrect m e if I’m w rong, but… N ow , dear reader, you probably… D oes this sound… to you? METADISCOURSE Figure 4: Ädel’s model of metadiscourse (Ädel, 2006: 38, adapted). 3. Some issues in metadiscourse theory In this section I will discuss a number of issues in metadiscourse theory. I have selected these issues for discussion, first, because they are important in themselves and, second, because I will return to them later in my discussion of SNs. 3.1. Metadiscourse as interpersonal or textual An issue that has already come up more than once in the above brief review of the literature is that of the relationship between the claimed interpersonal and textual functions of metadiscourse. As we have seen, early theoreticians such as Vande Kopple, divided metadiscourse according to these two types of meaning. However, as Ädel has noted, the understanding of the term “textual” departs from that of Halliday. For Halliday, the textual function is concerned with cohesion, theme/rheme and given/new. Vande Kopple’s “text connectives” would seem to fit in here, but the other categories she classifies as textual – code glosses, validity markers and narrators – seem to go beyond this. They seem to be more interpersonal, in fact. Validity markers (words like “perhaps”, “may”, “might”, “undoubtedly”), to take just one of these sub-categories, would be 03 IBERICA 29.qxp:Iberica 13 29/03/15 21:44 Página 21 3.1. Metadiscourse as interpersonal or textual An issue that has already come up more than once in the above brief review of the literature is that of the relationship between the claimed interpersonal and textual functions of metadiscourse. As we have seen, early theoreticians such as vande Kopple, divided metadiscourse according to these two types of meaning. however, as Ädel has noted, the understanding of the term “textual” departs from that of halliday. For halliday, the textual function is concerned with cohesion, theme/rheme and given/new. vande Kopple’s “text connectives” would seem to fit in here, but the other categories she classifies as textual – code glosses, validity markers and narrators – seem to go beyond this. They seem to be more interpersonal, in fact. validity markers (words like “perhaps”, “may”, “might”, “undoubtedly”), to take just one of these sub-categories, would be classified as markers of modality, according to halliday, and thus come under the interpersonal function. Another point made by Ädel (2006: 17) is that the textual and interpersonal functions are not at the same level. If the primary function of metadiscourse is to guide readers or listeners in how to interpret the propositional content of the text, then this is clearly interpersonal rather than textual. Ädel (2006: 17) points out that, in line with this view, mauranen (1993) argues that a textual function for metadiscourse is superfluous and that the fundamental function is interpersonal. As we have also seen, hyland joins mauranen and Ädel in this view and sees all metadiscourse as interpersonal. I will return to this issue very briefly in my discussion of SNs. 3.2. The size of the linguistic unit Because metadiscourse is a functional phenomenon, it goes without saying that there is no single formal linguistic category which can be identified as its mode of realisation. It can be noted from the examples given in the above literature review, that metadiscourse units may take the form of individual words, phrases, clauses or even whole series of clauses. Indeed, the unit of analysis tends to vary according to the analyst, some preferring larger units and some smaller. Ädel (2006: 41) suggests that smaller units are more useful if a researcher is interested in finding out the greatest possible differences and uses of metadiscourse across texts, while larger units of analysis are more useful if a researcher is interested in distinguishing between metadiscourse and non- metadiscourse across texts. Ädel and mauranen (2010) refer to the former JohN FLowERdEw Ibérica 29 (2015): 15-3422 03 IBERICA 29.qxp:Iberica 13 29/03/15 21:44 Página 22 approach as “thin” and the latter as “thick”. The thin approach, they argue, is more suited to quantitative methods and the thick approach is better suited to more qualitatively oriented methods. Ädel (2006: 52) also makes the point that, if the sentence is taken as the unit of analysis, then it is not possible to maintain the precept that metadiscourse is non-propositional, because there are bound to be propositions in whole sentences (or clauses). however, this is to assume that smaller units cannot carry propositional meaning, which is not the case. Another, more convincing, argument presented by Ädel, is that, in employing the larger unit, a whole clause or sentence is likely to carry with it more than one metadiscoursal function (a point I will return to in the next section on multi-functionality). For corpus-based approaches (e.g. hyland, 2005; Ädel, 2006), a further advantage of smaller units is that a much larger number of search terms can be employed (e.g. hyland, 2005, includes some 300 items in his inventory of metadiscourse items). with the corpus approach, however, one cannot be sure that a given item is always employed in its metadiscoursal function. As hyland states, “[i]t must be remembered, of course, that all items can realize either propositional or metadiscoursal meanings and that many can express either interactive or interpersonal meanings” (page 218). Based on this, hyland argues that every item should be studied in its sentential context. Although hyland may argue that every item should be treated in such a way, it is not clear if this is actually the case in his own work. This is not such a big problem if the aim of a study is to compare two corpora, provided that the same list of search terms is used in each. however, it does mean that no categorical statements can be made about the frequency of metadiscourse items in any given corpus. Claims can only be made in relative terms, comparing one (sub-)corpus to another. In the case of Ädel (2006), on the other hand, with the use of a smaller set of items than hyland’s 300 (she searched on subject forms of first and second person pronouns for her personal metadiscourse category and 61 words and lemmas for her impersonal category), makes it clear that the non-metadiscoursal items were eliminated from her counts (by manual analysis) and she is thus able to make clear-cut claims about the frequency of the metadiscourse items she includes in her searches. 3.3. The multi-functionality of metadiscourse items Another issue in metadiscourse research is the fact that items recognised as metadiscursive may perform more than one function at the same time. This REvISITINg mETAdISCouRSE Ibérica 29 (2015): 15-34 23 03 IBERICA 29.qxp:Iberica 13 29/03/15 21:44 Página 23 problem is greater the larger the unit of analysis, as already noted above. For example, in hyland’s model (see Table 2) “my purpose is noted” is given as an example of a frame marker; however, “my”, which is a part of this unit, is given as an example of an engagement marker. So there is clearly overlap here. To give another example, in Ädel’s model (Table 4), “my conclusion” is given as an example of the Personal – writer oriented category (presumably because of the “my”), but, although it is not given as an example, “conclusion” could also be classified as Impersonal –Text/Code-oriented (by analogy with the two examples given of “question” and “definition”). The solution adopted by Ädel (2006: 25) to the problem of multi- functionality is to decide on one of the possible functions as primary and to classify the item accordingly. This problem is less likely to arise, it can be pointed out, if single items are taken as the unit of analysis. 3.4. Representativeness in corpus-based approaches to metadiscourse The sampling procedures of Ädel and hyland referred to in the previous paragraph highlight the fact that such corpus approaches can only give a description of a well-defined sample of (potential) metadiscourse rather than covering the full range of metadiscourse in a given corpus. hyland (2005: 29), for example, states that he does not include indicators of affect or lexical evaluation (e.g. evaluative adjectives) in his model, because an operationalization of the model at such a level of delicacy would be impractical. Another problem with empirical studies of metadiscourse highlighted by hyland (2005: 30) is that of “insider opacity”, i.e. how discourse communities may have understandings of metadiscoursal features which are inaccessible to the analyst. This is particularly the case if the analysis is of academic discourse, which may involve a lot of “insider” language. Such language, of course, cannot be incorporated into corpus searches unless the analyst is helped by a specialist informant or informants. A further issue with corpus approaches is how to balance out the different categories in any model. If we take a taxonomy such as that of hyland (2005), for example, which has two major categories (interactive and interactional), with five sub-categories in each, an issue is how to avoid over- or under-representing a category in the searches. For example, in his inventory of search terms, hyland has far fewer code glosses than he does boosters. Presumably, these lists are derived from a word and/or clusters JohN FLowERdEw Ibérica 29 (2015): 15-3424 03 IBERICA 29.qxp:Iberica 13 29/03/15 21:44 Página 24 frequency list. But does the higher representation of one category as compared to another indicate that this category is more important or significant in metadiscourse terms than the other? That is to say, does it play a bigger role in communicating how the message is to be interpreted by the reader/listener? This issue is especially important when it comes to pedagogical application. As has been noted on many occasions (e.g. Swales, 2002), high frequency does not always mean high pedagogical priority2. To make a final point concerning the representativeness of corpus findings, I offer the following quotation from Ädel (2006: 48): A review of the literature on metadiscourse makes it clear that, for the most part, scholars do not mention methods of counting, let alone the difficulties involved. If we do not know how the findings of a study have been arrived at, it is difficult to have confidence in it; we do not know how representative it is. This is a basic issue in any form of academic research. In this respect, the models offered by hyland (2005) and Ädel (2006) are admirable in presenting inventories of the metadiscourse items used in their searches. 4. Metadiscourse and signalling nouns having reviewed some models of metadiscourse and considered some theoretical issues, as indicated in my introduction, I will now consider the notion of SN and how this category might fit into a model of metadiscourse. 4.1. The notion of signalling noun The notion of SN has its origins in work by winter (1977). winter identified three types of vocabulary which are important in establishing relations between clauses: type 1 consists of subordinators such as “although”, “except”, “unless”, and “whereas”; type 2 is made up of sentence connectors such as “as a result”, “however”, “indeed”, and “therefore”; and type 3 is made up of open class items3 – nouns, verbs and adjectives –that have the potential to connect clauses in a similar way to the type 1 and type 2 items. It is the nouns in this group that provided the inspiration for the notion of SNs (see also work by e.g. Francis, 1986, 1994; REvISITINg mETAdISCouRSE Ibérica 29 (2015): 15-34 25 03 IBERICA 29.qxp:Iberica 13 29/03/15 21:44 Página 25 Ivanič, 1991; and Schmid, 2001, among others for similar work). Although I will focus on SNs here, I may mention briefly that the other parts of speech (type 1 and type 2 vocabulary) have the potential of contributing towards metadiscourse. For example, winter’s type 1 and type 2 items correspond to hyland’s “transition markers” category of interactive resources, or stance markers (hyland, 2005: 49)4. SNs are abstract nouns like “fact”, “idea”, “possibility”, “problem”, and “result” which are non-specific in their meaning when considered in isolation and specific in their meaning by reference to their linguistic context. The following are some examples in the context of the clause or nominal group: (1) the fact is that we’ve all eaten genetically-modified crops (2) the hypothesis that the chemical beginnings of life can occur in atmospheres that are only mildly reducing (3) the point is not to go through all the steps (4) the possibility of curing sexual psychopaths (5) the situation where you’ve had to make a choice between two things In each of these examples, the SN (in bold) is realized, or specified, in the text which is underlined. SN specifics may also be found not in the same clause or nominal group as the SN, but in adjoining clauses, sentences, or even longer stretches of text, as exemplified in the following examples, where it is to be noted that the realization may either precede or follow the SN: (6) This geometry is probably favored because the trans configuration leaves a vacant low-ENERgY orbital that can be occupied by the two d electrons. According to this explanation… (7) Several factors have led price to become a more important component of marketing strategy. First, economic dislocations in the late 1970s and early 1980s made consumers more price-sensitive. Second, competition from lower-priced foreign imports has led American firms either to try to compete on a price basis, or to escape price competition by going after high-price segments. Third, companies have realized the benefits of segmenting markets by price and are offering high-, medium-, and low- priced brands in a product line. Finally, deregulation in many industries has led to greater price competition, increasing the importance of price. JohN FLowERdEw Ibérica 29 (2015): 15-3426 03 IBERICA 29.qxp:Iberica 13 29/03/15 21:44 Página 26 It is important to note that SN is a functional notion, not a formal one. Abstract nouns have the potential to act as SNs, but they are only counted as SNs if they are realized in the surrounding text, as in the above examples. Although unspecific in meaning in isolation, an SN nevertheless indicates how the piece of discourse to which it refers (referred to as its specifics, or its realization) is to be understood. It is a “fact”; it is a “hypothesis”; it is a “possibility”; it is a “point”, it is a “situation”; it is an “explanation”; it is (several) “factors”. The relations between an SN and its specifics are in fact complementary, each affecting the meaning of the other; the SN provides the specific meaning for the SN, but the SN indicates how the specifics are meant to be understood in relation to the surrounding discourse. The important point for the argument of this paper is that the SN performs a metadiscoursal function, in indicating how the stretch of text to which it refers is to be interpreted, whether this be in the same clause or across clauses. It may also be noted that SNs perform a textual function in allowing for the distribution of information across and within clauses, although space precludes a discussion of this issue. In my research with Richard w. Forest (Flowerdew & Forest, 2015), we have used three sub-corpora of lectures, textbook chapters and journal articles split further into 12 disciplines across the natural and social sciences. The corpus is referred to as the Flowerdew Corpus of Academic English (FCAE) and it consists of approximately 650,000 words. one of the most striking findings in this work is that SNs occur more than twice as frequently in the social sciences as they do in the natural sciences (the split is approximately 70/30). The reason for this is that natural sciences make greater use of technical terms, which reduces the need for SNs. The overall average frequency across the whole corpus is one SN per 37 words, an indication of the importance of this category, although there is considerable variation in frequency across disciplines and genres. The total count of different SN types in the FCAE is 845. we have classified SNs into six major semantic categories, as follows: • acts, which refer to material actions or events, e.g. “application”, “conversion”, “reaction” • locutions, which refer to verbal activity, e.g. “declaration”, “question”, “statement”; REvISITINg mETAdISCouRSE Ibérica 29 (2015): 15-34 27 03 IBERICA 29.qxp:Iberica 13 29/03/15 21:44 Página 27 • ideas, which refer to mental phenomena, e.g. “feeling”, “idea”, “thought”; • facts, which represent information about the world, “effect”, “result”, “thing”; • modal facts; which construe probability, usuality, obligatoriness, inclination, and ability, e.g. “duty”, “possibility”, “probability”; • circumstantial facts, which present information in terms of how, when where and why, e.g. “manner”, “period”, “way”. Classifying SNs in this way allows us to compare disciplines and genres in terms of these semantic categories. our findings are fairly consistent in ranking the semantic categories in the following order in terms of frequency, with little variation across genres and disciplines: fact, idea, circumstantial fact, locution, act and modal fact. This tells us, in very broad terms, that academic discourse in general is concerned with, first, representing the focus of the discourse as facts; then, with discussing ideas; next, with considering the circumstances of those facts, and ideas; and after that, with presenting the locutions (most probably by others) with regard to those facts. The functions of representing actions or facts about the world in terms of their probability are less frequent than the four others. one type of SN which, following, e.g. Francis (1986, 1994) and Ivanič (1991), we refer to as text nouns, nouns which refer to parts of the text, nouns like “section”, “paragraph”, and “paper”, is also worthy of mention with regard to the literature on metadiscourse, given that it is included in metadiscourse taxonomies (e.g. mauranen, 1993; hyland, 2005). one related feature of SNs that we have discovered in our corpus work is that many words which may function as SNs occur in prefabricated patterns, as adjuncts, e.g. “as a result”, “for example”, “in fact”, “in principle”. we have not counted examples such as these as SNs, because they have become grammaticalised, as adjuncts. Adjuncts such as these would fit into winter’s type 2 vocabulary rather than his type 3. They would also fit into some models of metadiscourse. They perform a similar function to SNs, but are not counted as SNs. JohN FLowERdEw Ibérica 29 (2015): 15-3428 03 IBERICA 29.qxp:Iberica 13 29/03/15 21:44 Página 28 5. Signalling nouns and metadiscourse Starting with vande Kopple’s metadiscourse category of “illocution marker”, which “make explicit to our readers what speech or discourse act we are performing at certain points in our texts” (page 84), most models of metadiscourse have something similar. hyland’s category “frame marker” refers to “discourse acts” (page 49) and Ädel (2006), has a category “metatext”, which “spells out the writer’s (and/or the reader’s) discourse acts” (page 36). In spelling out discourse acts, as should be clear from the above examples, we are in the territory of SNs. while vande Kopple presents only illocutionary verbs as examples of such markers, most such verbs have their nominal counterparts. Thus where vande Kopple gives the examples “hypothesize that”, “to sum up”, “we claim that”, and “I promise to”, we could equally well substitute nominal (SN) versions such as “my hypothesis is that”, “my conclusion is that”, “my claim is that”, or “my promise is that”. Two writers on metadiscourse do include nominal markers. They are Ädel, who includes “question”, “definition”, “discussion”, “example”, and “conclusion” as examples of her “metatextual” category of metadiscourse and hyland (2005), who includes the following nominals as frame markers: “aim”, “focus”, “goal”, “intention”, “objective”, and “purpose” (see appendix). Such nominals as these provided as examples by Ädel and hyland, it should be clear from the above section on SNs, are prime examples of potential SNs. however, Ädel only has five examples of SNs and hyland has just 6, while in the FCAE, as mentioned above, there is a total of 845 different SN types5. All of these SNs “spell out the writer’s discourse acts” to use Ädel’s terminology (page 36), so a strong case can be made for including them in any model of metadiscourse. The reason, I think, why vande Kopple and others focus on illocutionary verbs rather than their nominal counterparts is probably because their approach is primarily theoretical6. A lot has been written about illocutionary verbs (also referred to as “performatives”), at least since Austin (1962) and Searle (1969), while the literature on what we might refer to as the SN phenomenon has a much lower profile. one interesting model of metadiscourse that I have not reviewed above, but is of interest in this respect, is that of Beauvais (1969). The title of Beauvais’s contribution is “A speech act theory of metadiscourse” and his model takes as its starting point illocutionary verbs. he defines metadiscourse as “illocutionary force indicators that identify expositive illocutionary acts” (page 15). Although the starting point is illocutionary verbs and illocutionary acts, which are signalled REvISITINg mETAdISCouRSE Ibérica 29 (2015): 15-34 29 03 IBERICA 29.qxp:Iberica 13 29/03/15 21:44 Página 29 by such verbs (“expositive illocutionary acts”), Beauvais allows for “secondary expositive illocutionary acts”, which may include nominal counterparts of illocutionary verbs7. Nevertheless, the default category in Beauvais’s model, which is a theoretical one, not an empirical one, is still the illocutionary verb. If we take a corpus-driven approach and look in the FCAE, we find, in fact, that nominal (SN) forms are far more frequent than their verbal counterparts. “I/we hypothesise/hypothesize” does not occur at all in the FCAE (“we hypothesize” occurs once), while “hypothesis/hypotheses” occurs 90 times as an SN in the FCAE8. Similarly, there are no occurrences of “I/we claim”, while there are 77 instances of “claim” as an SN in the FCAE. To give a third example, there are again no instances of “I/we promise”, while “promise” as an SN occurs 21 times in the FCAE9. So the point I want to make in this section is that SN is a category that has been overlooked somewhat in models of metadiscourse and it might be taken into account more seriously in any theoretical or empirical approach to the topic. The question that follows is how would such a model be operationalized in a corpus-based study? Flowerdew and Forest (2015) provide an inventory of all of the SNs in the FCAE and this could form the basis for searching in other academic corpora10. one would probably not want to use all 845 SN types as search terms, but a selection of the most frequent could make up the inventory. In my introduction to the section of this article concerning issues in metadiscourse theory, I said that I would return to these issues in my discussion of SNs, which is what I will now do. So, first, let me turn to the issue of the textual and interpersonal functions. A distinctive feature of SNs is that both of these functions are important for SNs. Insofar as they indicate to the reader/listener what type of meaning to attach to a stretch of text, they can be considered to be interpersonal. But, insofar as they link up parts of a text they at the same time perform a textual function (see Flowerdew, 2003 for more on this). Next, let us consider the size of the unit of analysis. In the above discussion, for a number of reasons which I will not repeat here, I have suggested that smaller items may be better than larger items as the unit of analysis. with SNs, we obviously have a single item unit, so this is an argument in support of this approach. It makes the issue of counting much easier, with each unit being counted only once. Concerning the issue of multi-functionality, because we are dealing here with single-word JohN FLowERdEw Ibérica 29 (2015): 15-3430 03 IBERICA 29.qxp:Iberica 13 29/03/15 21:44 Página 30 items, this would seem to be an advantage and less of an issue. Finally, concerning the representativeness issue, if representativeness is the goal, in the 845 items from the FCAE, we have a fairly comprehensive list on which to base searches. however, other academic corpora, representative of other genres and disciplines, might contain other SN types not appearing in the 845 word list and the blind application of the FCAE list might not be appropriate. Needless to say, an application of the list to other non-academic corpora would complicate matters further. So this is an issue that needs addressing in any application designed for maximum representativeness. Researchers would need to search their corpus to identify additional SN types. But, of course, they could use the 845 word list as a stop list, which would be likely to cover most of the SN types occurring in the corpus, especially if it was an academic one. Space precludes further discussion of how SNs might be operationalized in a model of metadiscourse, but I hope to have given a flavour of how such a procedure might be put into practice. 6. Conclusion In this article I have reviewed a selection of models of metadiscourse and have discussed a number of important issues in metadiscourse theory: metadiscourse as interpersonal or textual; the size of the linguistic unit in metadiscourse research; the multi-functionality of metadiscourse items; and the issue of representativeness in corpus-based approaches to metadiscourse. I have then introduced the notion of SN, showing how the theory has been developed. Finally, I have argued that SNs represent an important resource for communicating intended meaning, i.e. that SNs are metadiscoursal, and that the category might be incorporated into models of metadiscourse. In my introduction, I situated this article in the context of the need to come up with optimal descriptions of genres as a basis for a genre-based pedagogy. This article has identified metadiscourse as an important dimension of such a description and has argued that SNs should be a part of such a description. metadiscourse, and within that SNs, can easily be introduced into a consciousness-raising model of genre-based pedagogy (Swales, 1990), where students can be invited to interpret, produce and critique their use (Pérez-Llantada, 2006: 82). Such an approach can make REvISITINg mETAdISCouRSE Ibérica 29 (2015): 15-34 31 03 IBERICA 29.qxp:Iberica 13 29/03/15 21:44 Página 31 learners aware of how metadiscursive patterns in the text can help to decipher the author’s intended message, can show how authors can guide readers/listeners to intended meanings and can indicate how a text hangs together, or coheres. At the same time, in terms of production, such an approach can empower learners to go beyond propositional meaning, introducing them to ways to make their intended meanings clear and signpost to their readers/listeners how to proceed through the text. Article history: Paper received 13 December 2014 Revised paper accepted 17 January 2015 References JohN FLowERdEw Ibérica 29 (2015): 15-3432 Ädel, A. (2006). Metadiscourse in L1 and L2 English. Amsterdam: John Benjamins. Ädel, A., & A. Mauranen (2010). “Metadiscourse: Diverse and divided perspectives”. Nordic Journal of English Studies 9(2): 1-11. URL: http://ojs.ub. gu.se/ojs/index.php/njes/issue/current [17/January/2015]. Austin, J.L. (1962). How to Do Things with Words. New York: Oxford University Press. Beauvais, P.J. (1989). “A speech act theory of metadiscourse”. Written Communication 6: 11-30. Biber, D. (1988). Variation across Speech and Writing. Cambridge: Cambridge University Press. Crismore, A. (1989). Talking with Readers: Metadiscourse as Rhetorical Act. New York: Peter Lang. Crismore, A., R. Markkanen & M.S. Steffenson (1993). “Metadiscourse in persuasive writing: A study of texts written by American and Finnish university students”. Written Communication 10(1): 39-71. Flowerdew, J. (2003). “Register specificity of signalling nouns in discourse” in C. Meyer & P. Leistyna (eds.), Corpus Analysis: Language Structure and Language Use, 35-46. Amsterdam: Rodopi Publishers. Flowerdew, J. (2011). “Action, content and identity in applied genre analysis for ESP”. Language Teaching 44(4): 516-528. Flowerdew, J. & R.W. Forest (2015). Signalling Nouns in English: A Corpus-based Discourse Approach. Cambridge: Cambridge University Press. Francis, G. (1986). “Anaphoric nouns”. Discourse Analysis Monographs 11. University of Birmingham. Francis, G. (1994). “Labelling discourse: An aspect of nominal-group lexical cohesion” in M. Coulthard (ed.), Advances in Written Text Analysis, 83-101. London: Routledge. Harris, Z. (1959). “The transformational model of language structure”. Anthropological Linguistics 1,1: 27-29. Hyland, K. (2005). Metadiscourse: Exploring Interaction in Writing. London: Continuum. Ivanič, R. (1991). “Nouns in search of a context: A study of nouns with both open- and closed-system characteristics”. International Review of Applied Linguistics 29,2: 93-114. Jacobson, R. (1960). “Closing statements: Linguistics and poetics, style in language” in T.A. Sebeok (ed.), Style in Language, 350-76. New York: Wiley. Noble, W. (2010). “Understanding metadiscourse use: Lessons from a ‘local’ corpus of learner academic writing”. Nordic Journal of English Studies (Special issue on metadiscourse) 9,2: 145-169. Pérez-Llantada, C. (2006). “Signaling speaker’s intentions in university lectures. Towards a phraseology of textual metadiscourse in academic speech” in C. Pérez-Llantada & G.R. Ferguson (eds.), English as a Glocalization Phenomenon. Observations from a Linguistic Microcosm, 59-88. Valencia: Prensas de la Universidad de Valencia. Schmid, H.-J. (2000). English Abstract Nouns as Conceptual Shells: From Corpus to Cognition. Berlin: Mouton de Gruyter. 03 IBERICA 29.qxp:Iberica 13 29/03/15 21:44 Página 32 John Flowerdew is a professor in the department of English, City university of hong Kong. his research and teaching is in the field of Applied Linguistics, focusing on discourse studies (including academic discourse analysis, critical discourse analysis, political discourse analysis and corpus-based approaches) and language education. he has authored or edited 14 books and special editions of journals and has published over 100 book chapters and internationally refereed journal articles. he serves on the editorial boards of a range of international journals and book series and is regularly invited to give plenary talks at conferences internationally. he is concurrently a visiting professor at the Centre for Corpus Approaches to Social Science (CASS) at Lancaster university. his latest book (with R.w. Forest) is “Signalling Nouns in English: A corpus-based discourse approach” (Cambridge university Press). NOTES 1 Space precludes a listing of many other examples of the narrower approach, but one example would be that of Noble (2006). 2 In corpus terms, the question of where priority should lie can be at least partially answered by work with learner corpora (e.g. Ädel, 2006; Noble, 2006). Such work can indicate where learners’ main difficulties lie. 3 winter claimed type 3 to be a closed set of some 100 items, but this claim has subsequently been refuted and the number of such items has been shown to be an open set (Flowerdew & Forest, 2015). 4 hyland (2005: 49) states: “Transition markers are mainly conjunctions and adverbial phrases which help readers interpret pragmatic connections between steps in an argument. They signal additive, causative and contrastive relations in the writer’s linking, expressing relations between stretches of discourse”. 5 Pérez-Llantada (2006: 73), in her study of metadiscourse in a spoken academic corpus, also identifies what she refers to as “phoric nouns” (examples she gives are “thing”, “question”, “point”, “issue” and “idea”) as realising metadiscursive functions. REvISITINg mETAdISCouRSE Ibérica 29 (2015): 15-34 33 Searle, J.R. (1969). Speech Acts. Cambridge: Cambridge University Press. Tognini-Bonelli, E. (2001). Corpus Linguistics at Work. Amsterdam and Philadelphia: John Benjamins. Vande Kopple, W.J. (1985). “Some explanatory discourse on metadiscourse”. College Composition and Communication 36: 82-93. Vande Kopple, W.J. (2002). “Metadiscourse, discourse, and issues in composition and rhetoric” in E. Barton & G. Stygall (eds.), Discourse Studies in Composition, 91-113. Cresshill, NJ: Hampton Press. Verschueren, J. (2004). “Notes on the role of metapragmatic awareness in language use” in A. Jaworksi, N. Coupland & D. Galasinski (eds.), Metalanguage: Social and Ideological Perspectives, 53-73. Berlin: Mouton de Gruyter. Swales, J.M. (2002). “Integrated and fragmented worlds: EAP materials and corpus linguistics” in J. Flowerdew (ed.). Academic Discourse, 150-164. London: Longman, Pearson Education. Williams, J. (1981). Style: Ten Lessons in Clarity and Grace. Boston: Scott Foresman. 03 IBERICA 29.qxp:Iberica 13 29/03/15 21:44 Página 33 6As opposed to using a “corpus-driven” approach and starting with discourse and then developing the theory (Tognini-Bonelli, 2001). 7 “Secondary acts do not use or imply first-person subjects; instead, they attribute the expositive act to someone other than the speaker/writer of the discourse or text” (Beauvais, 1989: 22). 8 To be fair, the passive form “hypothesized/hypothesized” does occur, 12 times. 9 The fact that nominal forms are more frequent than verbal ones is not surprising, given what we already know about academic discourse and its high frequency of nominalization (e.g. Biber, 1988). 10 Schmid (2000) also has an extensive inventory, which could be merged with that of Flowerdew and Forest (2015). JohN FLowERdEw Ibérica 29 (2015): 15-3434 03 IBERICA 29.qxp:Iberica 13 29/03/15 21:44 Página 34