Read doi:10.1016/j.pragma.2007.10.003 text version

Available online at

Journal of Pragmatics 40 (2008) 95­113

The pragmatic role of textual and interpersonal metadiscourse markers in the construction and attainment of persuasion: A cross-linguistic study of newspaper discourse

Emma Dafouz-Milne

Department of English Language and Linguistics, Faculty of Philology, Universidad Complutense de Madrid, Ciudad Universitaria, 28040 Madrid, Spain Received 24 November 2004; received in revised form 5 September 2007; accepted 5 October 2007

Abstract This paper seeks to explore the role that metadiscourse markers play in the construction and attainment of persuasion. In a cross-linguistic perspective, two elite newspapers, the British The Times and the Spanish ´ El Pais have been chosen, both because of their status and because of the political and rhetorical influence they exert in their respective national cultures. Based on the analysis of the textual and interpersonal markers found in a corpus of 40 opinion columns, 20 written in English and 20 in Spanish, this study aims to identify which metadiscourse categories predominate in this type of newspaper discourse and how they are distributed according to cross-cultural or cross-linguistic preferences. In addition, using a group of informants, this research has attempted to discover how metadiscourse operates as a persuasive mechanism in texts. Findings suggest that both textual and interpersonal metadiscourse markers are present in English and Spanish newspaper columns, but that there are variations as to the distribution and composition of such markers, specifically in the case of certain textual categories (i.e. logical markers and code glosses). Regarding the persuasive effect of metadiscourse, informants were in agreement that a balanced number of both textual and interpersonal markers was necessary to render the text persuasive and reader-oriented. # 2007 Elsevier B.V. All rights reserved.

Keywords: Newspaper discourse; Metadiscourse markers; Contrastive rhetoric; Persuasion; Corpus linguistics

1. Introduction This paper explores the role of textual and interpersonal metadiscourse markers in the construction and attainment of persuasion. Using a contrastive corpus of newspaper articles

E-mail address: [email protected] 0378-2166/$ ­ see front matter # 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.pragma.2007.10.003


E. Dafouz-Milne / Journal of Pragmatics 40 (2008) 95­113

written in English and Spanish, this study seeks to identify which metadiscourse markers characterise newspaper discourse, specifically opinion columns, and which of these markers are found to function more persuasively, according to the evaluation of a group of informants. The notion of metadiscourse will be adopted as an analytical framework since it has proved to be useful for textual analysis, agglutinating some of the explicit items that writers use to guide or direct readers through a text so both the text and the writer's stance is understood. Although the presence and function of metadiscourse markers has been examined in a number of different contexts, including textbooks (Crismore, 1984; Hyland, 1999, 2000), student writing (Crismore et al., 1993), science popularizations (Crismore and Farnsworth, 1990), advertisements (Fuertes-Olivera et al., 2001) and research articles (Mauranen, 1993; Luuka, 1994; Valero´ ~ Garces, 1996; Moreno, 1997, 1998; Hyland, 1998, 1999, 2000, 2001; Mur Duenas, 2007), surprisingly little attention (with the exception of Le, 2004) has been given to the genre of newspaper discourse and, to my knowledge, no study so far has contrasted metadiscourse in English and Spanish opinion columns (but see Dafouz, 2000, 2003). Newspaper discourse, and opinion columns in particular, can be considered ``some of the most adequate examples of persuasive writing in all countries, setting standards for written persuasion'' (Connor, 1996:143). Moreover, and given the wide audience they target and the fact that they are among the most widely read genres in newspapers (Bell, 1991; Reah, 1998), opinion columns serve to acquire and reinforce much of the readers' knowledge and beliefs (van Dijk, 1988). These texts, like editorials, deal with topics that are ``considered to be of particular societal importance at the time of publication'' (Le, 2004:688); however, unlike editorials, they are texts signed by a subject expert and may not reflect the official stance of the newspaper. Despite these texts' persuasive nature, the automatic acceptance of the ideas presented does not always occur. In order to persuade, columnists have to present the propositional material in a form that the potential audience will find most convincing and attractive. Furthermore, to succeed, writers need to create a credible textual persona or ethos and develop an appropriate attitude towards their readers and the claims they present. In the construction of this textual persona, metadiscourse plays a vital role. However, whilst the persuasive function of metadiscourse in a text has been sufficiently discussed and proved (see Mauranen, 1993; Hyland, 1998, 2005:63­71; Dafouz, 2003), the persuasive effect on an actual audience is not so evident. The present study, then, goes on to explore, using a questionnaire, whether metadiscourse markers actually attain persuasion on a group of selected informants and, if so, how such persuasion is metadiscursively articulated. In the following, a description of the taxonomy used in this analysis will be presented in section 2, section 3 provides a corpus description, while the general findings regarding the number of metadiscourse markers are presented in section 4. A discussion of the results for metadiscourse marker presence is presented in section 5, where section 5.1 deals with differences and similarities in the number and use of metadiscourse markers from a cross-linguistic perspective, whereas section 5.2 discusses their persuasive effects; finally, the section includes some comments from the informants involved in the research. 2. The role of metadiscourse markers in opinion columns: a textual and interpersonal taxonomy Although more than two decades have passed since the publication of some of the first works on metadiscourse (Williams, 1981; Crismore, 1984; Vande Kopple, 1985), interest in this notion from both a theoretical and practical perspective is still lively. Proof of this is the wealth of studies that have adopted a metadiscoursal approach (Crismore, 1989; Crismore et al., 1993; Mao, 1993;

E. Dafouz-Milne / Journal of Pragmatics 40 (2008) 95­113


~ Hyland, 1998; Dafouz, 2003, 2006; Le, 2004; Dahl, 2004; Mur Duenas, 2007; Neff and Dafouz, in press, to name a few), as well as the recent publication of Hyland's book (2005), which synthesises different conceptions, reviews the most relevant studies and aims to ``offer a more robust, explicit and useful model of metadiscourse'' (2005:x). Metadiscourse refers to those features which writers include to help readers decode the message, share the writer's views and reflect the particular conventions that are followed in a given culture. Following Hyland (2005:37), in this study metadiscourse is defined as ``the cover term for the self-reflective expressions used to negotiate interactional meanings in a text, assist the writer (or speaker) to express a viewpoint and engage with readers as members of a particular community.'' Metadiscourse thus is not simply a stylistic device, but is dependent on the rhetorical context in which it is used and the pragmatic function it fulfils (Mao, 1993:270). Explicitness is a key feature of metadiscourse since, in addition to being a practical means of identification and comparison, it represents the author's overt attempt to create a particular discoursal effect. Metadiscourse markers can be of many types and adopt various forms. They can range from a single word (`probably') to a full sentence (`the next point covered in this article deals with the topic of economy'), several sentences or even a whole paragraph. A number of taxonomies on metadiscourse markers have been proposed since initial interest began some decades ago (see Vande Kopple, 1985; Crismore et al., 1993; Beauvais, 1989; Hyland, 1998, 2005; Dafouz, 2003, inter alia). Most of these classifications (with the exception of Beauvais', 1989) generally organise the linguistic units under the functional headings of textual and interpersonal metadiscourse. Textual metadiscourse refers to the organisation of discourse, while interpersonal metadiscourse reflects the writer's stance towards both the content in the text and the potential reader. Recently, Hyland (2005) and Hyland and Tse (2004) have put forth a stronger interpersonal view on metadiscourse, claiming that all metadiscourse categories are essentially interpersonal since they need to take into account the readers' knowledge, textual experiences and processing needs. Thus, they proposed a change in the terminology adopting Thompson's (2001) label of interactive (instead of textual) and interactional (instead of interpersonal) metadiscourse. Although the present study aligns with the principle that metadiscourse categories are intrinsically interpersonal and ultimately aim to persuade the reader, we prefer to continue using the functional distinction of textual and interpersonal metadiscourse markers. This work, however, goes a step further in the functional division of textual and interpersonal metadiscourse, since it was found (Dafouz, 2003; Neff and Dafouz, in press) that on a functional level, Spanish and English newspaper articles share some findings in the type and even in the number of metadiscourse categories employed. Thus, in this paper the functions of metadiscourse categories as well as the linguistic form they adopt will be examined. This division will offer a variety of subcategories that will enable us to cover both the pragmatic functions of metadiscourse markers and the linguistic devices used to carry out such functions. The next sections offer a detailed account of the textual and interpersonal categories analysed in this study. 2.1. Textual metadiscourse Under this heading, seven categories have been included, as summarised in Table 1. The first of these categories refers to logical markers. These markers express semantic and structural relationships between discourse stretches, and help readers interpret pragmatic connections by explicitly signalling additive (and, furthermore. . . ), adversative (but, however. . . ), and conclusive relationships ( finally, in sum. . . ) in the text. Sequencers, the second category, mark particular positions in a series and serve to guide the reader in the presentation of different arguments in a


E. Dafouz-Milne / Journal of Pragmatics 40 (2008) 95­113

Table 1 Textual metadiscourse categories Macro-category Logical markers Express semantic relationships between discourse stretches Sequencers Mark particular positions in a series Reminders Refer back to previous sections in the text Topicalisers Indicate topic shifts Code glosses Explain, rephrase or exemplify textual material Parentheses Punctuation devices Reformulators Exemplifiers Illocutionary markers Explicitly name the act the writer performs Announcements Refer forwards to future sections in the text Subcategory Additive Adversative Consecutive Conclusive Examples and / furthermore/in addition/moreover. . . or / however / but. . . . so (as a result) / therefore / as a consequence. . . Finally / in any case. . . first / second / on the one hand,. . . on the other. . . .

Let us return to / as was mentioned before. . . .

in political terms / in the case of the NHS. . . . When (as with the Tories now) . . . Tax evasion: it is deplored in others but not in oneself. in other words / that is/to put it simply. . . for example / for instance. . . . I propose / I hope to persuade. . .

there are many good reasons/as we'll see later. . . .

particular order (in the first place, secondly). Reminders refer back to previous sections in the text in order to retake an argument, amplify it or summarise some of the previous argumentation (as was mentioned before). Topicalisers explicitly indicate some type of topic shift to the reader so that the argumentation can be easily followed (regarding the health problem/as for X). Code glosses explain, rephrase, expand or exemplify propositional content. Overall, they reflect the writer's expectations about the audience's knowledge or ability to follow the argument (that is, in other words, for instance). In this analysis, we have also included parentheses (following Hyland, 1998:443) and colons as instances of glosses, since much of the reformulation and exemplification that takes place in opinion columns is implemented through these visual markers. These non-verbal signals, along with others (such as underlining, capitalization, italics, etc.), are regarded and classified as visual metadiscourse (a term put forth by Kumpf, 2000), which shows their importance in the analysis of text. Illocutionary markers explicitly name the act the writer performs though the text (I hope to persuade, I back up this idea. . . ). Announcements refer forward to future sections in the text in order to prepare the reader for prospective argumentation (as will be seen below). 2.2. Interpersonal metadiscourse Five main categories comprise this section, as Table 2 shows. Firstly, hedges refer to markers that withhold full commitment to the statements displayed in the text. From a linguistic point of

E. Dafouz-Milne / Journal of Pragmatics 40 (2008) 95­113 Table 2 Interpersonal metadiscourse markers Macro-category Hedges Express partial commitment to the truth-value of the text Certainty markers Express total commitment to the truth-value of the text Attributors Refer to the source of information Attitude markers Express writers' affective values towards text and readers Commentaries Help to establish reader-writer rapport through the text Subcategory Epistemic verbs Probability adverbs Epistemic expressions Examples May / might / it must be two o'clock Probably / perhaps / maybe It is likely Undoubtedly / clearly / certainly


`x' claims that. . . / As the Prime Minister remarked Deontic verbs Attitudinal adverbs Attitudinal adjectives Cognitive verbs Rhetorical questions Direct address to reader Inclusive expressions Personalisations Asides Have to / we must understand / needs to Unfortunately / remarkably / pathetically It is absurd / it is surprising I feel / I think / I believe What is the future of Europe, integration or disintegration? You must understand, dear reader We all believe/let us summarise What the polls are telling me / I do not want Diana (ironically for a Spencer) was not of the Establishment

view, epistemic verbs (may, might, would), probability adverbs ( perhaps, maybe) and epistemic expressions (it is likely, it is probable. . . ) have been analysed. Certainty markers, by contrast, express full commitment to the statements presented by the writer (undoubtedly, of course, naturally. . . ). Attributors perform a double function in the text: they mention explicitly the source of the information (as the Prime Minister indicated), while at the same time using these references of authoritative value with persuasive goals. Attitude markers express the writer's affective values towards the reader and the content presented in the text. Linguistically, these markers can adopt the form of deontic verbs (must, have to. . . ), attitudinal adverbs (surprisingly. . . ), adjectival constructions (it is difficult, imposible. . . ) and cognitive verbs (I think, I believe. . . ). Finally, commentaries help to establish and maintain rapport with the audience by means of rhetorical questions (is this the right attitude?), direct appeals (dear reader, you), personalisations (I, me, my feelings) and asides (Diana -ironically for a Spencer-). 3. Corpus description and procedure The corpus1 on which this study is based comes from the opinion columns of two quality ´ newspapers: The Times, from the UK and El Pais, from Spain. The texts, a total of 40 (accounting

Part of this corpus, specifically The Times subcorpus, has also served as data for the SPICLE research team (the Spanish section of ICLE Corpus, the International Corpus of Learner English directed by Prof. Sylviane Granger in Louvain and Dr. Jo-Anne Neff in Spain). This project consists of the compilation of a written corpus of learner English around the world for comparative purposes with other learner corpora and expert writing (see Granger and Petch-Tyson, 2003).



E. Dafouz-Milne / Journal of Pragmatics 40 (2008) 95­113

Table 3 Texts chosen for persuasive effect English TextEng01 TextEng02 TextEng03 TextEng04 TextEng05 TextEng06 Spanish TextSpa01 TextSpa02 TextSpa03 TextSpa04 TextSpa05 TextSpa06 Total metadiscourse 11 17 38 41 63 78 Total metadiscourse 15 17 48 51 63 73 Textual metadiscourse 6 9 18 22 21 30 Textual metadiscourse 10 11 20 36 28 44 Interpersonal metadiscourse 5 8 20 19 42 48 Interpersonal metadiscourse 5 6 28 15 35 29

for 46,815 words), have a similar length (approximately 1000 words each) and were matched for topic in order to ensure comparability. The articles cover the following topics: Economic affairs, European Union, International affairs (Israel and Palestine) and issues dealing with National Health Service and Education. All texts were stored electronically (FilemakerPro), and in addition to automatic searches, the analysis was also carried out manually to ensure its validity. As was stated before, since the type and appearance of metadiscourse categories are extremely varied and many metadiscourse categories are multifunctional, a context-sensitive analysis of each marker had to be carried out before it was finally classified. In addition, the fact that there are very few metadiscourse studies dealing with the Spanish language (with the exception of Valero´ Garces, 1996; Moreno, 1997, 1998) made the analysis of the Spanish data more complex. The statistical analysis employed non-parametrical means (Mann­Whitney test), since most of the data in the sample were not normally distributed. The Mann­Whitney test was chosen as a conservative measure of differences between the English and Spanish data to compare two sets of data, based on their ranks below and above the median. Results have been standardised to a common basis (mean per 1000 words) to compare the frequency of occurrence. For the analysis of the persuasive effect, 12 opinion columns from the original 40 were selected (listed in Appendix A). We based this choice on the number of metadiscourse markers that each text contained and on the topic covered.2 Consequently, the corpus for persuasion was finally comprised of six texts in English and six texts in Spanish: two texts in English and two in Spanish with a high proportion of metadiscourse markers, two with a medium proportion, and two with a low proportion. An additional reason for reducing the original corpus was found in the principles of feasibility and availability, since informants could only analyse a limited number of texts without the task becoming too demanding. Table 3 summarises this information. To avoid aprioristic influences on the informants, the texts were edited, omitting the name of the author and newspaper and the title of the article. The readers were given a questionnaire in

2 It is important to control the topic variable since as other studies have shown (Crismore and Farnsworth, 1990; Hyland, 1999; Thompson, 2001; Dafouz, 2003) the topic covered may affect the type, frequency and distribution of the metadiscourse markers used.

E. Dafouz-Milne / Journal of Pragmatics 40 (2008) 95­113


order to rank the texts from 1 to 5, 1 representing the least persuasive text and 5 the most persuasive. They were also invited to explain their decisions by writing open comments. The questionnaire (see Appendix B) was loosely adapted from Connor (1987) and Connor and Lauer's (1988:146) model for persuasive writing and contained three basic criteria for evaluating persuasive effect: rational appeals, credibility appeals and affective appeals. Rational appeals are logical lines of reasoning: arguments based on the structure of reality, offering argumentation by example, illustration and model, and analogy and metaphor. Also comparisons, facts and statistics, and cause and effect examples fall into this category. Credibility appeals include the writer's personal experience, knowledge of the subject, and awareness of the audience's values. In other words, devices such as personal pronouns and personal references to build a credible textual persona are used here. And finally, affective appeals include the use of concrete and charged language, of vivid pictures, and of metaphors to evoke emotion and sentiment in the audience. After prior piloting of the questionnaire with a small group of teachers to ensure comprehensibility, the texts were sent to the English Departments of seven different universities in the area of Madrid.3 The evaluators addressed were native speakers of the target languages involved and specialists in the teaching of English or Spanish as a foreign language. Of the 300 questionnaires originally distributed, 67 were returned, 32 from the Spanish informants and 34 from the English. From a quantitative perspective means were calculated for each of the 6 texts ranked. From a qualitative perspective, it was found that the informants' written comments regarding their views on persuasive texts were numerous and essential to the evaluation of the texts. For this reason, some of these comments have been included in the discussion section. 4. Findings 4.1. Findings for metadiscourse markers On a general level, the quantitative analysis reveals that the texts written in Spanish used a higher number of textual metadiscourse markers than did the texts in English (Spanish, n=496 , English, n=334). As for interpersonal markers, the Spanish texts used a lower number of metadiscourse markers than the English texts did (Spanish, n=331 and English, n=424). However, from a statistical perspective, these differences were not significant4 ( p = 0.357), a result that endorses the principle that metadiscourse markers act cross-linguistically and are present in many different genres. In the case of this particular genre, opinion columns, the presence of metadiscourse items confirms the importance of this resource to construct persuasive texts while following certain genre conventions. A detailed look into the subcategories that comprise the textual and interpersonal taxonomy reveals, however, interesting differences from a cross-linguistic angle (see Figs. 1 and 2). Within textual metadiscourse, the presence of logical markers and code glosses is very different in both corpora. While the Spanish texts abound in the use of additive markers (n = 111) to link ideas, the English texts prefer the use of adversative markers (n = 82) to construct arguments. The numerical preponderance of logical markers over the other metadiscourse markers responds to organisational

´ The universities contacted were: Universidad Complutense de Madrid, Universidad Autonoma de Madrid, ´ Universidad de Alcala (for the Spanish professors) and the international programmes of St. Louis University, Middlebury College and Syracuse University in Madrid (for the English professors). 4 Results have been standardised to a common basis (mean per 1000 words) to compare the frequency of occurrence.



E. Dafouz-Milne / Journal of Pragmatics 40 (2008) 95­113

Fig. 1. Results for textual metadiscourse markers.

principles and the clarification of ideational connections. However, there are other possible rhetorical interpretations to this finding that will be presented in the discussion section. Regarding code glosses, findings display statistically significant differences ( p = 0.0016) with the Spanish writers using these explanatory devices much more frequently than do their English counterparts. Parentheses, for instance, were used three times more in the Spanish data (n = 108) than in the English (n = 35) and occurrences of reformulators (of the type that is, in other words. . . ) in Spanish double the English usage in Spanish double the English usage (n= 16 vs n= 6).

Fig. 2. Results for interpersonal metadiscourse markers.

E. Dafouz-Milne / Journal of Pragmatics 40 (2008) 95­113


As for the rest of the textual markers, the analysis reveals that sequencers were also very numerous in the Spanish texts, with 65 instances versus 6 in the English data. However, on closer scrutiny it was found that the use of sequencers was not evenly distributed among texts, with two authors hoarding most of the instances. The remaining categories (i.e. reminders, topicalisers, illocutionary markers and announcements) displayed a low frequency of occurrence in the texts surveyed, both in English and Spanish. Tentatively, one might suggest that the presence of these items is not necessary, given the short length of these articles and the linguistic economy that writers have to exercise in their columns. From these findings, it seems that our writers choose to include other textual markers (i.e. logical markers, code glosses) which may play a more decisive role in the intended interpretation of the text. Regarding interpersonal markers, findings disclose that hedges were the most frequently used category in both sets of groups: English (n = 180) and Spanish (n = 121). In fact, hedges were the most numerous of all the markers analysed in the corpus, even more so than logical markers. This finding seems to confirm the crucial importance of combining fact and mitigated opinion in newspaper discourse in order to attain effective persuasion. Moreover, cross-culturally, it reveals that both Spanish and English writers follow parallel rhetorical conventions in the articulation of persuasion by means of metadiscourse hedges. Linguistically speaking, findings reveal that modal epistemic verbs are the most frequently used strategy to express caution, and that the Spanish writers employ expressions similar to those used by the English writers to protect their claims. Specifically, the Spanish writers alternate the use of poder/can (n = 37) with parecer/ seem (n = 27), while the English opt for would in the first place, followed by may, could, can and seem. Altogether, epistemic verbs account for 141 tokens of the total number of hedges used in this corpus, a finding that shows that this is the linguistic device preferred by newspaper writers to hedge their opinions. In diminishing order, epistemic adverbs follow this list of hedging resources, with the English group using a slightly higher proportion (n = 28) than the Spanish ´ (n = 16) and in both cases mostly grouped under two adverbs perhaps/quizas and probably/ probablemente. Finally, tokens for other epistemic expressions are rather low, with only 15 instances between the two groups, the English writers utilizing the expression is/are likely on 8 occasions. After hedges, attitudinal markers were the second most frequent marker with both corpora showing a very similar number. On closer examination, the study reveals that deontic verbs are the most frequent resource in both groups with 122 instances; however, the English writers employ these verbs on 75 occasions while the Spanish only use them 47 times. The uses revolve ´ around the verbs should/deberia and must/deber. Other deontic verbs found are have to/hay que. In addition to deontic verbs, other attitudinal expressions of the type it is + evaluative adjective were also present. A total of 57 of these constructions appeared in the corpus, with very similar findings for both sets of writers (English n = 32, Spanish n = 25). The most common cluster in this category coincides in both groups (but their linguistic articulation and perspective are different). Thus, while in English the most frequent expression was it is difficult in the Spanish ´ case it was no es facil/it is not easy. Lastly, cognitive verbs of the type I believe/think/suspect/ creo/opino/sospecho. . . were also found in the data with no differences among the groups, ´ ´ followed by attitudinal adverbs like ironically/ironicamente, incredibly/increiblemente, desgraciadamente/unfortunately, with only 11 instances in the whole corpus. Certainty markers occupy the third place in terms of frequency of occurrence, with Spanish and English texts showing an almost identical number of tokens (Spanish n = 44; English n = 43). Within the linguistic preferences, our English data show that the expression of course is the most numerous, closely followed by indeed and surely. In the Spanish data, the constructions es cierto


E. Dafouz-Milne / Journal of Pragmatics 40 (2008) 95­113

(or cierto es)/certainly dominate with 20 instances, and sin duda/no doubt and desde luego/of course follow. Finally, commentaries display the lowest scores within the interpersonal markers. Rhetorical questions show the highest proportion of instances with a total of 73 (English n = 47 and Spanish n = 26). At a considerable distance, there are imperative constructions with 15 examples in English and only 4 in Spanish, which suggests that such expressions are not favoured by the Spanish group. Plural expressions (us, our, we) account for 15 tokens while there are only nine examples of personalisations (I, my, me) in both groups. By contrast, there is quite a numerous representation of the markers known as `asides' in the data (n = 58), a finding that suggests that opinion columns tend to use this resource to insert explicit opinions and appraisals without having to articulate them into full sentences and also allowing the reader to interpret them more freely. The discussion section will offer examples for the major findings found in this study. 4.2. Findings for the questionnaire on persuasive effect The 67 informants that participated in the evaluation of the 12 opinion columns generally chose as more persuasive the texts with a balanced number of metadiscourse markers, in other words, texts where there was a medium index of metadiscourse, both textual and interpersonal. Figs. 3 and 4 below display the findings. Curiously, the lowest scores (3,0 by the English teachers and 2,9 by the Spanish) were given to texts with a low index of metadiscourse markers, although these scores were not too different from the ones obtained by the texts with a high index (3,8 and 4,3), a result that suggests that, given the task, informants believed all the texts to be somewhat inherently persuasive. Focusing on the metadiscursive composition of each text, in order to ascertain which markers predominated (whether textual or interpersonal) the analysis revealed that the 12 texts chosen

Fig. 3. Metadiscourse markers and their persuasive effect in English corpus.

E. Dafouz-Milne / Journal of Pragmatics 40 (2008) 95­113


Fig. 4. Metadiscourse markers and their persuasive effect in Spanish corpus.

shared similar types of metadiscourse categories, predominatively logical markers and code glosses in the textual function, and hedges and attitudinal markers in the interpersonal one. This result reproduces the same tendencies in the selected corpus as in the whole corpus (i.e. the 40 texts), with a balanced number of textual and interpersonal markers in most texts. Moreover, the four texts that obtained the highest scores all included a higher proportion of hedges than of attitudinal or certainty markers; a finding that can be interpreted as writers favouring persuasion by means of identification and negotiation with the audience rather than by imposition. 5. Discussion 5.1. Metadiscourse markers The general findings from this study reveal that metadiscourse markers play a key role in the construction of persuasion in opinion columns. Moreover, they suggest that metadiscourse is an important feature of professional rhetorical writing in languages other than English and that the categories that comprise it, so widely used in the English-speaking contexts, are also applicable in the Spanish milieu. As seen in Fig. 1, the findings showed that, in regard to textual categories, logical markers clearly stand out as the most frequently used items, with Spanish writers preferring additive markers and English writers favouring adversative ones. A possible explanation for this difference could be linked to the way in which different communities view and construct their argumentation. Thus, while in the Spanish tradition, argumentation is very frequently built by adding positive warrants to the thesis statement, always moving in the same direction, in the English tradition arguments normally follow a dialectical approach (i.e. pros and cons), a strategy that necessarily implies the use of adversative markers. This finding matches contrastive research


E. Dafouz-Milne / Journal of Pragmatics 40 (2008) 95­113

by Mauranen (1993) which claims that Finns, like our Spanish group, build their argumentation using a progressive strategy that entails moving forward in the establishment of ideas and adding evidence to the original claim. In her study, the Anglo-American writers, by contrast, exhibit a retrogressive strategy, based on the reconstruction of an argument using the pros and cons of an opinion. The two examples below extracted from the data exhibit how the English author builds his argument on the Israeli-Palestinian conflict using a contrastive approach and consequently employing the adversative marker but. By contrast, the Spanish author constructs his rationale listing a set of reasons that head in the same direction by means of additive markers: and/y, ´ ´ moreover/ademas and also/tambien. (1) (. . . ) TextEng03 ``What has the new Jerusalem in store?'' The Arab order has always looked precarious, but it has shown an amazing capacity for survival, different as the regimes are. (. . . ) Mr Arafat probably agrees that the talks should be accelerated but he is essentially a bargainer. He will not expect to get everything he wants at once, but further talks, with Israel under pressure, could give him a negotiating advantage. ´ TextSpa03 ``Un paseo por Hebron'' [A walk through Hebron] Los acuerdos de paz de Oslo y los pasos que se dieron rompieron por fin el status ´ ´ quo y demostraron que era posible lo que parecia imposible. (. . . ) Ademas estos ´ ´ ´ acuerdos mostraron que habia una corriente flexible y pragmatica y tambien desvelaron que estaban dispuestos a hacer las concesiones necesarias para lograr la paz. [The Oslo peace agreements and the steps taken finally broke the status quo and proved that what seemed impossible was possible. (. . . ) Moreover these agreements showed that there was a flexible and pragmatic trend and they also revealed that they were ready to make the necessary concessions to achieve peace].

(2) (. . . )

Regarding the high number of code glosses in both sets of texts, these results suggest that writers are aware of the broad audience they are addressing and consequently they are also aware of the need to include a number of explicit reading cues as well as more exemplifications, thus ensuring that the text is read as intended. Specifically, the use of parentheses by the Spanish writers ( p = 0.0016) puts forward the idea that Spanish opinion columns may exhibit a greater freedom to include what the English-speaking rhetorical principles consider ``supplementary or digressive'' material. Interestingly, this finding coincides with Kaplan's (1966) influential study of textual organisation across languages and cultures, in which he pointed out that Romance languages (Spanish included) tolerate `digressions' better than Anglo-Saxon languages and construct argumentative texts incorporating information that may be not strictly pertinent in the eyes of a non-Spanish writer. The question that arises here is to what extent should the information that appears within parentheses be considered of primary or secondary importance and whether there may be cultural differences in this interpretation. Overall, it seems that the use of code glosses, and specifically of parentheses, helps the writer introduce information (whether relevant or not) in a fairly reduced space, offering facts that in the view of the columnist are necessary, anticipating the needs of readers, while at the same time adjusting to the tight space restrictions of this newspaper genre. As an example, in the fragment below the journalist includes information within parentheses that is pertinent for the understanding of the British attitude towards the issue of belonging or not to Europe. It seems that the use of this resource, rather than minimising the

E. Dafouz-Milne / Journal of Pragmatics 40 (2008) 95­113


information in brackets, emphasizes it, giving it more saliency, so that the reader actually notices its importance. (3) TextSpa01 ``Contra Europa'' [Against Europe]

~ ~ En las recientes campanas electorales europeas (Gran Bretana y Francia) pocos han defendido la Europa de Maastrich. (. . . ) Ni Blair, ni Cook (Ministro de Asuntos Exteriores) son thatcherianos. ´ ´ ´ Precisamente porque heredan un pais con una dramatica factura social (sanidad, educacion, ´ ´ ´ asuntos socials. . . ) no pueden aceptar una Europa cuya union (economica) no esta vinculada a los derechos sociales. [In the recent European polls (Great Britain and France) few have defended the Maastricht Europe (. . . ) Neither Blair nor Cook are thacherites. Precisely because they are inheriting a country with a dramatic social debt (health, education, social affairs. . . ) they cannot accept a Europe whose union (economic) is not linked to social rights]. Curiously, these findings for code glosses are contrary to Le's (2004) expectations, where she states that in comparison to academic writing, newspaper writing (and editorials in particular) needs to present the information with less complexity since it is targeted to a wider audience. Thus, in her view, the presence of these reformulation and exemplification devices is not required. From the data analysed in this paper, what I have found is that the use of explicit reformulation and clarification markers (of the type in other words, that is. . . ) is indeed low, but that authors exemplify and reformulate by means of other more economic and compact devices such as the aforementioned parentheses and colons. As mentioned in the findings section (4.1), the remaining textual markers (specifically, announcements and reminders) are virtually non-existent in this corpus; the interpretation being that the prospective and retrospective functions of these items are not necessary in such a shortlength genre. In addition, the reduced amount of topicalisers found suggests that authors prefer to introduce their topics and subtopics in the text without using overt markers, probably because the main topic is clearly stated in the title (see Appendix A) or because the number of topics covered in a column is normally limited. In regard to interpersonal markers, hedges stood out as the most frequent category, confirming their crucial role in persuasive texts, where the writer needs to strike a difficult balance between commitment to his/her ideas and respect and dialogue with the reader. In other words, by means of this feature writers can anticipate possible opposition to their claims (by expressing statements with precision but also with caution and modesty), while simultaneously, enabling the reader to follow the writer's stance without the writer appearing too assertive. These quantitative results coincide with other studies where hedges also hold a predominant position, irrespective of the genre and the languages analysed. Hedging, for example, has come to be seen as a key characteristic of academic discourse (Hyland, 1998; Moreno, 1998), be it in economic texts ~ (Moreno, 1998; Mur Duenas, 2007), medical research papers (Myers, 1989; Salager-Meyer, 1994), or advertisements (see Fuertes-Olivera et al., 2001), where indirectness is highly valued for different reasons. On the subject of the linguistic resources employed to hedge, the analysis suggests that both languages favour the use of epistemic verbs (i.e. can, could, would (or equivalent conditionals)/ poder, seems/parece) in detriment of other strategies. Again, this finding matches other studies where modal verbs are the most prototypical realisation of hedging. In our data would is the most frequently used modal (n = 45) followed by may (n = 28) could (n = 23) and can (n = 17). In Spanish, poder (n = 37) followed by conditional forms (functionally equivalent to would) account for 32 tokens. In a study on Spanish newspaper discourse, Romero-Gualda (1993)


E. Dafouz-Milne / Journal of Pragmatics 40 (2008) 95­113

observes that conditional forms are prototypical of media discourse since they enable the user to express an opinion or state a fact without a full compromise and allowing some room for discussion. Epistemic adverbs and constructions generally follow a similar trend, showing no substantial differences cross-linguistically. Certainty markers are an intrinsic characteristic of opinion columns, since readers expect to find the writer's opinion overtly stated. In these two corpora, certainty units, used in conjunction with modals, were found to favour persuasion, rather than expressing certainty. These forms seem to be used to create a sense of solidarity with readers when discussing issues that are in fact divisive. Given the relatively high number of attitudinal markers found in the data, it can be said that the writer's personal feelings (agreement or disagreement) prove to be a persuasive tool in the eyes of the reader. It seems, thus, that the key to an effectively persuasive text is the artful combination of weakening expressions (i.e. hedges) and strengthening ones (i.e. certainty markers and/or attitudinal markers) with the final intention of producing a discourse that is neither too assertive nor too vague. Again linguistically, the most frequent attitudinal expressions correspond to ´ deontic verbs in both groups (i.e. must/deber, should/deberia) with the English using nearly double the amount that the Spanish writers use. For the rest of the categories, there are no visible distinctions, a finding that indicates that both sets of writers favour the use of verbal expressions to signal the degree of importance granted to a particular statement and the effect with which they want it to be interpreted. Finally regarding commentaries, the most noticeable feature is the profusion of rhetorical questions (n = 45 in total) and asides (n = 51 in total) in both sets of texts. In my view, the two dominating principles for the use of these resources are interaction with the reader and solidarity with his/her processing of the text. However, Thompson (2001:61) suggests a more manipulative goal in the use of rhetorical questions when he states: ``the writer spells out the question that the cooperative reader expects to be answered and thus encourages the reader to accept the direction the text is taking.'' In other words, the reader just follows the path set out by the writer. In addition, the fact that rhetorical questions are multifunctional and introduce new topics in a text while concurrently address the reader, also makes their presence in texts highly useful. Personal markers, expressions of inclusion, and especially asides, also have a notable presence in the data. The use of these devices reflects the tendency of opinion columns to express opinion in a much more personal way than is the case for editorials or academic papers, for example. In fact, readers are usually searching for the explicit signalling of an author's personal stance when they read these articles, since one of the roles of opinion columns is to reveal a writer's individual thoughts and beliefs. 5.2. Persuasion and the presence of metadiscourse markers The informants' response on persuasion revealed that the texts with a balanced number of metadiscourse markers were regarded as the most persuasive. In general, it seems that readers prefer to be given some metadiscursive categories, both textual and interpersonal, to guide them through the text, rather than having to reconstruct and reinterpret the text without any explicit signposting. At the same time, readers regard as more persuasive those texts in which metadiscursive items are not over-represented and where there is still room for individual interpretation, opinion or disagreement. Interestingly, interpersonal metadiscourse is not the only determinant for persuasion; textual metadiscourse in the form of logical markers, which add, sequence, contrast or conclude a number of ideas is essential to the overall persuasive effect of a text. Equally essential are code glosses that include relevant information by rephrasing, offering examples and explaining further,

E. Dafouz-Milne / Journal of Pragmatics 40 (2008) 95­113


or sequencers that list a number of arguments so that the reader can follow the text comfortably. All in all, textual metadiscourse, although not overtly persuasive, fulfils a persuasive function and attains a persuasive effect. In the case of interpersonal markers, our informants agreed that the combination of hedges and attitudinal markers, together with the inclusion of certainty markers and personalisers, contribute to the development of a relationship with the reader. A relationship that, ultimately, may convince or not but that is inherently persuasive. Admittedly, in the questionnaire the interviewees were not asked to analyse the texts by focusing on any metadiscourse classification. Their guidelines were to evaluate the article holistically in terms of the global persuasive effect it had on them and the rational, credibility and affective appeals found. It was an impressionistic assessment, and many other variables obviously intervened when the interviewees had to decide on the degree of persuasion. Nevertheless, from the written comments of the informants some of the criteria followed to evaluate the persuasive effect of the texts could be traced. Concerning the two texts that received the highest scores, some of the most common remarks were the following: Inf-Eng03: Adequate choice of language and theme. Inf-Eng01: Very well and fully argued, anticipating counterarguments; good examples. Inf-Eng19: Serious, austere presentation backed up with factual data. Inf-Eng22: Includes the reader by using ``us'', ``we'' or ``our'' so that the reader is pulled by what the author is saying. ´ Inf-Spa17: Ejemplos concretos para ir a la base de la argumentacion. [Concrete examples taking us to the core of the argumentation] ´ ´ Inf-Spa19: La exposicion de ideas es clara y ordenada. La numeracion ayuda bastante. [The exposition of ideas is clear and structured. The use of numbers helps considerably]. Inf-Spa25: Bien estructurado e informativo. [Well-structured and informative]. These comments agree on the importance of well-structured texts and on solid argumentations and on interpersonal resources as key factors to persuasive discourse. Interestingly, there is explicit mentioning of the importance of some metadiscursive categories (obviously under a different term) to render the text persuasive. For instance, comments by informant 22 note the importance of certain interpersonal markers to enhance author-reader interaction (us, we, our), and informant 19 highlights the usefulness of numbers (or sequencers) to present an argument. By contrast, the two texts with the lowest scores were said to exhibit the following traits: Inf-Eng25: Subjective and poorly documented. Inf-Eng29: Use of vocabulary is extreme or exaggerated. Too many metaphors (. . . ) Use of the first person makes the article less objective and professional. Inf-Eng32: Too chatty and disjointed. Hard to follow due to the lack of organizational devices and the author's assumptions of what the reader already knows. ´ Inf-Spa13: Poco ordenado y muy denso. Falta de consideracion hacia un lector poco familiarizado. [Poorly organised and very dense. Lack of consideration towards not too well informed reader]. In these texts, the absence of metadiscourse markers organising and facilitating the reading task is found to be responsible for the lack of persuasion. In addition, informant 29 mentions that the use of certain personalisation strategies (first person singular) diminishes textual credibility. By and large, these comments should help us interpret the quantitative analysis of persuasion, but they should also remind us of the complex nature of the phenomenon studied and the impossibility of


E. Dafouz-Milne / Journal of Pragmatics 40 (2008) 95­113

separating the textual from the extra-textual factors that come into play when talking about persuasion. The following comment by one of the informants summarises exactly this idea: Inf-Eng16: Very-well written. Carefully organised and using lots of examples and figures. However, I found it very hard to detach my reactions from what I already knew I felt about the subject in hand. I am afraid I am not convinced at all! 6. Conclusions This paper has presented quantitative and qualitative research on the presence and persuasive function of metadiscourse markers in a corpus of Spanish and English opinion columns. Regarding similarities, the study reveals that textual and interpersonal metadiscourse markers are present both in the English and Spanish texts (although there are variations as to the distribution and composition of such markers). These similarities can be attributed to the newspaper-genre characteristics of opinion columns that seem to transcend the national culture and exhibit a certain uniformity across languages. Concerning differences, this study suggests that there is some room for internal variation across languages in the construction of opinion columns. For instance, the findings for logical markers underline a possible prospective pattern in the Spanish argumentation patterns in comparison to the retrospective pattern of the English argumentation patterns. Likewise, differences in the amount of code glosses, that is, in the type of supplementary material that English and Spanish writers consider to be relevant, are also worthy of note. From the linguistic composition of the textual metadiscursive markers, this study has discovered that there are no statistically significant differences in the way Spanish and English opinion columns articulate these items. And although certain preferences have been identified (such as the higher presence of deontic modality in English, especially must, or the lack of imperative expressions in the case of the Spanish corpus), there is an overriding cross-linguistic homogeneity. As regards the analysis of persuasion, the findings suggested that texts with a balanced number of textual and interpersonal metadiscourse items were considered as the most persuasive by our informants, followed by texts with a high number of these markers. The texts considered to be less persuasive of all by our informants were those with a low index of metadiscourse markers. In other words, it seems that our readers highly value texts that guide and show consideration towards the audience, by establishing a dialogic tenor without resulting too assertive or patronising. There are two important limitations in this study of persuasion: (a) the small-scale nature of the research and (b) the dynamicity and extra-textual dimension of the notion of persuasion itself. There are various extra-textual reasons not analysed here that play a decisive role in the judgment of persuasion (the topic, the type of newspaper, the section in the newspaper or the columnist). Thus, further studies could offer more insight into persuasion by using a larger corpus, focusing on different disciplines and genres and maybe involving other subjects. For instance, it would be interesting to contrast teacher evaluations' with that of novice readers in order to learn whether their view of a persuasive text turns out to be the same. Transferring these findings to pedagogical grounds, I believe there is an evident need to include metadiscourse markers, specifically interpersonal ones, in L1 and L2 writing courses (see Hyland, 2005:175­193). So far, only a limited range of metadiscourse markers (mainly textual) have been included in EFL texts or been taught with any regularity, and very rarely are they presented as playing a pragmatic and persuasive role in the interaction between writer, reader and text. Whilst experienced readers/writers may understand that both reading and writing

E. Dafouz-Milne / Journal of Pragmatics 40 (2008) 95­113


are context-rich, situational, constructive acts, many learners see reading and writing as merely an information-exchange process. Thus, helping students move beyond this simple, ideational view to a more complex, interpersonal model should be a teaching priority. Acknowledgments I would like to thank Dr. JoAnne Neff and two anonymous reviewers for their constructive criticisms and suggestions on earlier drafts of this article. My gratitude extends to Dr. Jacob Mey, the chief editor, for his support and encouragement. Appendix A. List of opinion columns selected for the persuasion task Spa01 Spa02 Spa03 Spa04 Spa05 Spa06 Eng01 Eng02 Eng03 Eng04 Eng05 Eng06 Contra Europa [Against Europe] Europa tortuga o cangrejo? [Europe, turtle or crab?] ´ Un paseo por Hebron [A walk through Hebron] ~ Cambio en la Gran Bretana? [Change in Great Britain?] Sanidad Cautiva [A captive Health System] Diana o la caja de los truenos [Diana or the powder keg] The Tories, Europe and the People What is the future of Europe, integration or disintegration? What has the new Jerusalem in store? Europe is rooting for Blair Closing the Health and Education gaps Now law could have shielded her ?

Appendix B. Sample questionnaire How to fill in the questionnaire 1. Read each of the 6 texts selected. 2. After reading them, rank the texts from 1 to 5 (5 = highest score and 1 = the lowest) according to the degree of persuasion that, in your opinion, the texts present. Please check the descriptors included below as reference model. 3. Explain the reasons for your decision. Thank you very much for collaborating. Descriptors 5 = The text is highly persuasive. The argumentation is very well-presented, the text is very well-structured and the author uses plenty of examples, facts and figures, to endorse the ideas presented (i.e. rational appeals). The author uses his/her personal experience and subject knowledge very appropriately to convince the audience (i.e. credibility appeals) and includes vivid language and emotional strategies (direct address to the reader, inclusive pronouns. . . ) in a very effective manner to evoke sentiment in the readership (i.e. affective strategies).



E. Dafouz-Milne / Journal of Pragmatics 40 (2008) 95­113





= The text is very persuasive. The argumentation is well-presented and well-structured. The author uses some examples, facts and figures, to endorse the ideas presented (i.e. rational appeals). The author uses his/her personal experience and subject knowledge to convince the audience (i.e. credibility appeals) and includes vivid language and emotional strategies (direct address to the reader, inclusive pronouns. . . ) to evoke sentiment in the readership (i.e. affective strategies). = The text is fairly persuasive. The argumentation is rather loose, the text is loosely structured and the author uses very scarce examples, facts and figures, to endorse the ideas presented (i.e. rational appeals). The author uses his/her personal experience and subject knowledge in a limited way to convince the audience (i.e. credibility appeals) and includes few effective samples of vivid language and emotional strategies (direct address to the reader, inclusive pronouns. . . ) to evoke sentiment in the readership (i.e. affective strategies). = The text is more or less persuasive. The argumentation is not well-presented, the text is not well-structured and the author does not use examples, nor facts or figures to endorse the ideas presented (i.e. rational appeals). The author does not use adequately his/her personal experience and subject knowledge to convince the audience (i.e. credibility appeals) and fails to include vivid language and emotional strategies (direct address to the reader, inclusive pronouns. . . ) to evoke sentiment in the readership (i.e. affective strategies). = The text is not persuasive. There is no argumentation in the text, no structuring of the text and the author does not use examples, nor facts or figures to endorse the ideas presented (i.e. rational appeals). The author does not use his/her personal experience and subject knowledge to convince the audience (i.e. credibility appeals) and fails to include (or uses inappropriately) vivid language and emotional strategies (direct address to the reader, inclusive pronouns. . . ) to evoke sentiment in the readership (i.e. affective strategies).


Beauvais, Paul. J., 1989. A speech act theory of metadiscourse. Written Communication 6, 11­31. Bell, Allan, 1991. The Language of News Media. Language in Society, vol. 17. Blackwell, Oxford. Connor, Ulla, 1987. Argumentative patterns in student essays: cross-cultural differences. In: Connor, U., Kaplan, R. (Eds.), Writing Across Languages: Analysis of L2 Text. Addison-Wesley, Reading, pp. 57­71. Connor, Ulla, 1996. Contrastive Rhetoric. Cross-cultural Aspects of Second Language Writing. Cambridge University Press, New York. Connor, Ulla, Lauer, Janice, 1988. Cross-cultural variation in persuasive student writing. In: Purves, A.C. (Ed.), Writing across Languages and Cultures. Sage, Newbury Park, CA, pp. 138­159. Crismore, Avon, 1984. The rhetoric of textbooks: metadiscourse. Journal of Curriculum Studies 16 (3), 279­296. Crismore, Avon, 1989. Talking with Readers: Metadiscourse as a Rhetorical Act. Peter Lang, New York. Crismore, Avon, Farnsworth, Rodney, 1990. Metadiscourse in popular and professional science discourse. In: Nash, W. (Ed.), The Writing Scholar. Sage, Newbury Park, California, pp. 118­136. Crismore, Avon, Markannen, Raija, Steffensen, Margaret, 1993. Metadiscourse in Persuasive Writing. A study of texts written by American and Finnish University students. Written Communication 10 (1), 39­71. ´ ´ Dafouz, Emma, 2000. El Metadiscurso como estrategia retorica en un corpus de textos periodisticos. Estudio contrastivo ~ de lengua inglesa y lengua espanola. Unpublished Ph.D. dissertation, Universidad Complutense de Madrid. [Metadiscourse as a rhetorical strategy in a corpus of newspaper articles. Contrastive study of English and Spanish texts.] Dafouz, Emma, 2003. Metadiscourse revisited: a contrastive study of persuasive writing in professional discourse. Estudios Ingleses de la Universidad Complutense 11, 29­52.

E. Dafouz-Milne / Journal of Pragmatics 40 (2008) 95­113


´ Dafouz, Emma, 2006. Estudio de los marcadores interpersonales en el comentario periodistico: estrategias para la ´ identificacion autor-lector en el texto [A study of interpersonal markers in newspaper columns: strategies for author~ ¨i reader identification]. Revista Espanola de Lingu´stica Aplicada 19, 67­82. Dahl, Trine, 2004. Textual metadiscourse in research articles: a marker of national culture or of academic discipline? Journal of Pragmatics 36 (10), 1807­1825. ´ ´ ´ ~ Fuertes-Olivera, Pedro-A., Velasco-Sacristan, Marisol, Arribas-Bano, Ascension, Samaniego-Fernandez, Eva, 2001. Persuasion and advertising English: metadiscourse in slogans and headlines. Journal of Pragmatics 33, 1291­1307. Granger, Sylviane, Petch-Tyson, Stephanie (Eds.), 2003. Extending the Scope of Corpus-Based Research. New Applications, New Challenges. Rodopi, Amsterdam. Hyland, Ken, 1998. Persuasion and context. The pragmatics of academic discourse. Journal of Pragmatics 30, 437­455. Hyland, Ken, 1999. Talking to students: metadiscourse in Introductory coursebooks. English for Specific Purposes (1), 3­ 26. Hyland, Ken, 2000. Disciplinary Discourses: Social Interactions in Academic Writing. Longman, London. Hyland, Ken, 2001. Humble servants of the discipline? Self-mention in research articles. English for Specific Purposes 20 (3), 207­226. Hyland, Ken, 2005. Metadiscourse. Exploring Interaction in Writing. Continuum, Oxford. Hyland, Ken, Tse, Polly, 2004. Metadiscourse in academic writing: a reappraisal. Applied Linguistics 25 (2), 156­177. Kaplan, Robert, 1966. Cultural thought patterns in intercultural communication. Language Learning 16, 1­20. Kumpf, Eric P., 2000. Visual metadiscourse: designing the considerate text. Technical Communication Quarterly 9 (4), 401­424. Le, Elisabeth, 2004. Active participation within written argumentation: metadiscourse and editorialist's authority. Journal of Pragmatics 36, 687­714. Luuka, Minna Riita, 1994. Metadiscourse in academic texts. In: Gunnarsson, B.L., Linell, P., Nordberg, B. (Eds.), Text and talk in Professional Context. ASLA (The Swedish Association of Applied Linguistics), Uppsala, pp. 77­88. Mao, Luming R., 1993. I conclude not: toward a pragmatic account of metadiscourse. Rhetoric Review 11 (2), 265­289. Mauranen, Anna, 1993. Cultural Differences in Academic Rhetoric. Peter Lang, Frankfurt am Main. Moreno, Ana Isabel, 1997. Genre constraints across languages: causal metatext in Spanish and English RAs. English for Specific Purposes 16 (3), 161­179. Moreno, Ana Isabel, 1998. The explicit signalling of premise-conclusion sequences in research articles: a contrastive framework. Text 18 (4), 545­585. ~ Mur Duenas, Pilar, 2007. A contribution to the intercultural analysis of metadiscourse in business management research ´ articles in English and Spanish: a corpus-driven approach. Unpublished Doctoral Dissertation. Universidad de Leon. Myers, Greg, 1989. The pragmatics of politeness in scientific articles. Applied Linguistics 10 (1), 1­35. Neff, JoAnne, Dafouz, Emma, in press. Argumentation patterns in different languages: an analysis of metadiscourse markers in English and Spanish texts. In: Putz, M., Neff, J. (Eds.), Developing Contrastive Pragmatics: Interlanguage and Cross-Cultural Perspectives. Mouton, Berlin. Reah, Deah, 1998. The Language of Newspapers. Routledge, London. Salager-Meyer, Francoise, 1994. Hedges and textual communicative function in medical English written discourse. ¸ English for Specific Purposes 13 (2), 149­170. ´ ´ ~ Romero-Gualda, Maria Victoria, 1993. El espanol en los medios de comunicacion. Arco/libros, Madrid. Thompson, George, 2001. Interaction in academic writing: Learning to argue with the reader. Applied Linguistics 22 (1), 58­78. ´ Valero-Garces, Carmen, 1996. Contrastive ESP rhetoric: metatext in Spanish-English economic texts. English for Specific Purposes 15 (4), 279­294. van Dijk, Teun, 1988. News as Discourse. Lawrence Erlbaum, Hillsdale, NJ. Vande Kopple, William, 1985. Some exploratory discourse on metadiscourse. College Composition and Communication 36, 82­93. Williams, Joseph M., 1981. Style: Ten Lessons in Clarity and Grace. Harper Collins Publishers. Emma Dafouz-Milne is associate professor at the Universidad Complutense de Madrid (Department of English Language and Linguistics), where she teaches Applied Linguistics. Her main research interests include contrastive rhetoric, written discourse analysis and content and language integrated learning (CLIL). She has published a number of articles on written discourse analysis with John Benjamins, Peter Lang and Rodopi, and works on CLIL in the International Journal of Bilingualism and Bilingual Education. She holds a Ph.D. from the Universidad Complutense de Madrid and a MSc. in Applied Linguistics from Edinburgh University.



19 pages

Report File (DMCA)

Our content is added by our users. We aim to remove reported files within 1 working day. Please use this link to notify us:

Report this file as copyright or inappropriate


Notice: fwrite(): send of 214 bytes failed with errno=104 Connection reset by peer in /home/ on line 531