Engaging the Congregation: The Place of Metadiscourse in Contemporary Preaching

Abstract

Most forms of contemporary preaching elevate the role of the listeners, who are seen as co-constructors of the sermon. This article investigates how preachers may respond to a call from homiletics to ‘turn-to-the-listener’ through their deployment of metadiscourse, operationalized as a framework for understanding preaching not primarily as gospel proclamation, but as a form of social and communicative engagement. A quantitative analysis of 150 sermons from three well-established English Christian denominations finds that metadiscourse is a prolific feature in preaching, with almost one word in 10 performing a metadiscursive function, most prominently by recourse to interactional metadiscursive resources. A qualitative analysis shows that, across the three denominations, preachers adopt metadiscursive practices that may serve to address expectations in the turn-to-the-listener preaching paradigm. The findings will inform further applied analysis of religious discourse, and will add constructively to discussions within homiletics, not least by suggesting how empirical and applied approaches to language can be of service to homiletics and the teaching of preaching.

INTRODUCTION

This study addresses a new context for research on interpersonally oriented language in the form of so-called ‘metadiscourse’ ( Hyland 2005 ) as well as a gap in the analysis of religious discourse by investigating contemporary preaching, and how metadiscourse may be of service to preachers from three well-established English denominations. Metadiscourse involves linguistic expressions, the main purpose of which is either: (i) to help guide listeners through the sermon by adding to the overall cohesion and coherence of the sermon, or (ii) to help involve listeners in the argument made in the sermon, principally by expressing a preacher stance or projecting variable points of view (cf. Hyland 2005 ). The highlighted metadiscursive expressions in sermon examples (1) and (2) illustrate these basic purposes of metadiscourse.

• (1) The Christian understanding of Resurrection therefore needs to be better explained.

• (2) If the only piece of evidence were Peter’s word, then, perhaps , Jesus’ resurrection could be dismissed.

At the heart of the investigation is this ‘principal research question’: What is the place of metadiscourse in contemporary preaching where so much emphasis is placed on the listener? More specifically, the investigation will establish: (i) how frequent metadiscourse is in preaching; (ii) whether certain forms of metadiscourse are particularly prevalent in preaching; and (iii) whether preachers’ use of metadiscourse may be conducive to turn preaching to the listener. Focusing on these issues will further our knowledge about how preaching is rhetorically shaped and reaches out to listeners. This research will also advance knowledge within the discipline of homiletics by adding crucial empirical and applied perspectives on sermonic language use, specifically by drawing attention to the communicative potential of linguistic metadiscourse.

Over a period of ∼40 years, and much influenced by ideas advanced in postmodern hermeneutics and homiletics, the nature and direction of Western European and North American Christian preaching has changed gradually but significantly. Long gone, Craddock ( 2001 : 13) argues, are the days when preachers representing established denominations ‘ascended the pulpit to speak of eternal certainties, truths etched forever in granite of absolute reality, matters framed for proclamation, not for discussion’. The homiletics literature talks of a paradigmatic shift from traditional argument-centered and deductive preaching at listeners to what is commonly known as ‘New Homiletic’ or ‘turn-to-the-listener’ preaching, typically shaped as a sermonic conversation with listeners ( Craddock 2001 ; Wesley Allen 2010 ). Sermons have changed from being concerned primarily with biblical hermeneutics to a kind of congregation-centered hermeneutics, the overarching purpose of which is to persuasively engage with listeners and move them toward subjectively applied understanding and to affect individual experience ( Rose 1997 ; Hogan 1999 ; Craddock 2001 ; Immink 2004 ), all the while making sure that listeners are able to explore the full spectrum of the confessional-interpretative space ( Wesley Allen 2010 ). Table 1 identifies some common denominators of this contemporary type of preaching.

Table 1:

Common denominators of New Homiletic preaching—the list of principles is not meant to be exhaustive

	Preaching ‘turned-to-the-listener’ typically:
Preaching Principle 1	Elevates listeners’ role in preaching; listeners are seen as co-constructors of the sermon and conversational partners.
Preaching Principle 2	Opens up an interpretative space and acknowledges listeners’ right to form their own opinions regarding claims made in the sermon.
Preaching Principle 3	Avoids a voice of authority; confirms faith more than it proves faith.
Preaching Principle 4	Has a sermon design that helps listeners to engage with the sermon during preaching.
Preaching Principle 5	Focuses on preaching as embodiment (experience) and performance.
Preaching Principle 6	Progresses inductively, as a coherent narrative, on the basis of lived experience about faith.

Source : Rose 1997 ; Craddock 2001 ; Gibson 2005 ; Wesley Allen 2010 .

Contemporary preaching is consequently a communicative event motivated by interpersonal ambitions and characterized by extensive use of interpersonally oriented language [highlighting a linguistic ‘interpersonal metafunction’ used to ‘enact social relationships’ as understood by Halliday ( 1994 : 34)]. However, as noted by, for example, Wuthnow ( 1992 : 60), our collected knowledge about ‘the ways in which religious discourse is actually put together’ is limited, and research in linguistics concerned with preaching is scarce.

PREVIOUS WORK: THE LINGUISTIC STUDY OF SERMONS

Calls from homiletics scholars for more research into sermons from other disciplines (e.g. Guthrie 2007 ) have largely been neglected by linguists (exceptions include: Dzameshie 1995 ; Cheung 1999 ; Koncar Bizjak 2008 ; Ethelston 2009 ; Johnstone 2009 ; Capone 2010 ). The only other study to date concerned with the metadiscourse of preaching is Boggel (2009) , whose investigation is firmly grounded in historical linguistics. Boggel’s diachronic account points to how different types of metadiscourse (what she refers to as intertextual -, text-organizing -, instructional -, and stance marking metadiscourse ) are a means for preachers to ‘guide their audience through their texts and persuade them of their points of view’ ( Boggel 2009 : 41). Her study highlights significant differences in the way the different types of metadiscourse are employed during different historical periods, suggesting that preachers’ deployment of metadiscourse is sensitive to contextual and societal factors.

On account of Boggel’s study we have some knowledge about how metadiscourse may be conducive to preachers’ addressing their interpersonal objectives, but her findings are representative of a period when reaching out to an audience was of little significance because of the nature and standing of preaching at the time. It is problematic that we hardly know anything about how contemporary preachers exploit metadiscourse during a time when the interpersonal appeal is absolutely essential for connecting with sermon listeners. The next section introduces the theoretical framework used in this study to investigate to what extent and how contemporary preachers use metadiscourse to engage their listeners.

METADISCOURSE AS A FRAMEWORK FOR UNDERSTANDING CONTEMPORARY PREACHING AS SOCIAL AND COMMUNICATIVE ENGAGEMENT

This study adopts Hyland’s model of metadiscourse, which is widely attested in applied linguistics research and operates with a taxonomy of metadiscourse categories that have proved to be productive for more than a decade ( Hyland 2005 ). Metadiscourse is operationalized as a framework for understanding preaching not primarily as persuasive proclamation, but as a form of social and communicative engagement between preacher and congregation, and it assumes an awareness on the part of the preacher that the sermon is a discourse which imposes certain social and communicative constraints on both preachers and listeners, for example, with regard to preacher–listener status, power, and authority (cf. Hyland 2005 ).

The highly dynamic model of metadiscourse adopted by Hyland (2005) is reflective of decision making during communication and consequently highlights the ‘system of making meanings’ ( Halliday 1994 : xvii), that is, language: ‘the decisions we make when interacting with others [and] choices motivated by intentions to express certain meanings in specific situations’ ( Hyland 2005 : 17). The ‘specific’ situation studied in this article is preaching.

Hyland’s model of metadiscourse has primarily been used to research metadiscourse in academic discourse contexts ( Hyland 2005 ). Recent research ( Malmström under review ) has shown that in many respects sermonic discourse is quite unlike academic discourse, whether spoken or written; this, then, would be a reason to question whether it is appropriate to use Hyland’s model for analyzing sermons. However, adoptions and adaptations of Hyland’s model for the study of other kinds of non-academic spoken as well as written (meta) discourses suggest that its application is by no means limited to written academic and scientific discourse [see, e.g. Cavalieri 2011 on the role of metadiscourse in counsels’ questions in the court room; Dafouz-Milne (2008) on opinion columns in newspaper discourse, or Jensen (2009) on discourse strategies in professional e-mail negotiations]. Indeed, there is no reason why Hyland’s basic assumption that text can be seen as social engagement between producer and audience should be restricted to academic/scientific discourse. To this end, adopting Hyland’s model to research metadiscourse in sermons could also be seen as a response to Ädel’s ( 2010 : 94) call for research into ‘the status and place of metadiscourse in non-academic language’.

According to Hyland ( 2005 : 37) ‘[m]etadiscourse is the cover term for the self-reflective expressions used to negotiate interactional meanings in a text, assisting the writer (or speaker) to express a viewpoint and engage with readers as members of a particular community’. In addition to this definition, Hyland ( 2005 : 38) poses three key principles for metadiscourse:

• Metadiscourse is distinct from propositional aspects of discourse. ¹

• Metadiscourse refers to aspects of the text that embody writer [speaker]–reader [listener] interactions.

• Metadiscourse refers only to relations which are internal to the discourse.

The use of the expression ‘first’ in (3) and (4) illustrates how these principles can be applied to distinguish between metadiscourse and non-metadiscourse.

• (3) There are various options as to how we can respond to that. First , we can embrace it and hone it and polish it and bring it to a fine art. Second, we can just potter on with our comforting, familiar, domestic half-truths.

• (4) First the women were afraid. Then they overcome their fears and ran […].

The basic metadiscursive function of ‘first’ in (3) is to index stages in the sermonic discourse, to alert the addressee that what follows is the first of several propositions forming part of a chain of hypothetical propositions. In this case the word ‘first’ is conducive to the listener’s navigation of the discourse and its inclusion is therefore metadiscursively motivated. The use of ‘first’ in (4) is non-metadiscursive as it refers to the first stage in a sequence of events in a narrative, clearly external to the sermonic discourse, and it has no bearing on the interpersonal relationship between preacher and listeners. The speaker simply asserts that something happened in a sequenced manner and ‘first’ is part of the asserted proposition.

Hyland (2005) makes a principled distinction in his model between interactive and interactional metadiscourse, as illustrated in Table 2 .

Table 2:

Metadiscourse resources

Category	Function	Examples
Interactive resources	Help guide the listener through the discourse
Transitions	Express semantic relation between utterances	in addition, but, thus, and
Frame markers	Refer to discourse acts, sequences, or discourse stages	finally, to conclude, my purpose here is to, first, second
Endophoric markers	Refer to information in other parts of the discourse	noted above, see fig x, in section 2
Evidentials	Refer to source of information from other texts/discourses	According to X, (Y, 1990), Z states
Code glosses	Help listeners grasp functions of ideational material	namely, e.g., such as, in other words
Interactional resources	Help involve the listener in the argument
Hedges	Withhold speaker’s full commitment to proposition	might, perhaps, possible, about
Boosters	Emphasize force or speaker’s certainty in proposition	in fact, definitely, it is clear that
Attitude markers	Express speaker’s (non-epistemic) attitude to proposition	unfortunately, I agree, surprisingly
Engagement markers	Explicitly refer to or build relationship with listener	consider/note that, you can see that, you, we (inclusive)
Self-mentions	Explicit reference to speaker	I, we (exclusive), my, our (exclusive)

Source : Hyland and Tse 2004 : 169.

The argument pursued in this article, on the basis of this framework of metadiscourse, is that different types of interactive and interactional metadiscourse are productive during preaching to help preachers turn sermons to listeners. A priori, it is assumed that interactive metadiscourse help preachers to facilitate listeners’ structural and logical perception, understanding and subsequent interpretation of a sermon utterance, and the way that utterance contributes to the overall cohesion and coherence of the sermon. The interactional dimension of metadiscourse is instead characterized by preachers’ concern to involve listeners in the sermon (i.e. the content/idea rather than the discourse, or content rather than form), sometimes by siding with listeners, emphasizing inclusiveness and shared beliefs, and sometimes by reinforcing that a position is the preacher’s own, a position which may be different from the one occupied by the listeners.

MATERIALS AND METHODS

This quantitative and qualitative study is based on an analysis of 150 sermon manuscripts (preached sometime between 2005 and 2011) from three well-established Christian denominations in England: the Church of England (nominally Protestant Anglican but with large variations across the spectrum), the Baptist Church, and the Roman-Catholic Church, totaling 189,239 words. Although small in corpus terms, this is an indicative sample of contemporary preaching. The three churches were selected in order to get a reasonably representative spread across Christian denominations and preaching traditions of different size; several sources reporting on religious affiliations in England (e.g. Peach 2005 ) estimate affiliation with the Church of England at ∼60 percent of all English Christians, the Roman-Catholic Church at ∼13.5 percent and the Baptist Church at <2 percent. In total, 50 Easter sermons (Palm Sunday through to Easter Monday) from each denomination were included, using only a single manuscript from any one preacher. Table 3 presents a summary of the data samples used.

The starting point for the ‘quantitative analysis’ was a list of linguistic elements commonly assumed to function metadiscursively provided in Hyland (2005) . This list is based on the occurrence of the linguistic elements in written academic discourse and words/phrases that did not occur at all in the sermon samples ( n = 45) were taken off the list. A close reading of three sermons from each of the three denominations resulted in the addition of a small number of words/phrases ( n = 27) to the list. In the end an amended list with 285 word/phrase types formed the basis for the search for metadiscourse in the sermon samples. Operating with pre-established categories and elements of metadiscourse obviously means that there may be linguistic elements potentially performing a metadiscursive function in the data that are not covered by the analysis. Consequently, I stress that this survey of metadiscourse in sermonic discourse is not to be considered comprehensive in the sense that all linguistic elements that could possibly perform a metadiscursive function are identified. Just as for Hyland and Tse ( 2004 : 175), the objective of this study is to investigate ‘different patterns of occurrence of metadiscourse’ and to give a first indicative picture of the status and place of metadiscourse in sermonic discourse.

AntConc ( Anthony 2013 ), a simple and freely downloadable concordance tool, was used to generate concordances for each of the 285 words/phrases appearing on the amended list. This was done for all three denomination samples. The concordance searches using the listed words resulted in a total of 18,216 returns. The returns were exported to three different Excel files. Subsequently, the concordance data had to be ‘cleaned’ to ensure that all returns included were actually instances of metadiscourse. A second phase involved coding all instances in the denomination samples according to metadiscourse categories (deciding whether something counts as a booster, a hedge, etc.). The principles for cleaning and coding the data [the defining criteria proposed by Hyland (2005) included above, and the general metadiscourse category descriptions in Table 2 ] were confirmed by a colleague from another discipline. Each of us looked at the same 600 concordance lines (randomly selected by taking every 30th instance in the combined sample) involving examples of most of the different metadiscursive categories listed in Table 2 . In identifying those instances to be deleted (because they were not metadiscourse) we agreed in 94 percent of the cases. In coding the remaining instances, we agreed in our categorization 89 percent of the time. This was thought to be sufficient to confirm the principles for analysis. The subsequent, complete, category coding by the author was done manually, item-by-item, and line-by-line by using simple tagging functions in the three Excel files. Although extremely time-consuming, it was thought to be worth the effort to guarantee correct coding of all items.

As the sermon samples are of different sizes, all frequencies reported have been normalized to occurrences/1,000 words to allow for comparisons across the three denomination samples.

To understand what discourse functions metadiscursive resources perform in preaching and how these address the Preaching Principles in Table 1 , a more qualitative approach had to be added to the investigation. A ‘qualitative analysis’ involved two steps, intended to be mutually supportive of each other. As a first step, 100 metadiscursive elements from each category of metadiscourse, or as many elements as were available (e.g. 100 elements performing a hedging function) were analyzed individually. For each occurrence, I asked the question: ‘In this context, what is the preacher’s rationale for using this metadiscourse item?’ Surrounding context was considered to add to the interpretation. Notes were taken throughout the process and eventually a tentative pattern of interpersonal functions emerged for each of the metadiscursive resources.

Table 3:

Overview of sermonic data used in this study

	Anglican	Baptist	Catholic	All churches
Number of sermon texts	50	50	50	150
Total no. of words (tokens)	64,439	76,751	48,049	1,89,239
Mean length of sermons	1,289	1,535	961	1,262

To better understand the interpersonal choices preachers make with respect to their metadiscourse, and to confirm or refute my interpretations regarding metadiscourse function, I then supplemented the analysis of metadiscursive functions with interview data. Twenty-one of the preachers who had contributed sermons were contacted and asked whether they would be willing to discuss language use in preaching. In the end 11 preachers (at least three from every denomination) responded positively and they were interviewed face to face in England in their respective home parishes/equivalent in October 2012. The format for the interview is best described as ‘semi-structured’ ( Drever 2003 ); the 11 preachers were asked open-ended questions about preaching practices in general and their own practices in particular. They were also asked to comment on particular instances of language use in their manuscripts. Finally, we discussed some of my interpretations. Following the interviews I revisited my notes taken during the qualitative analysis in the first step and reviewed some of the sermon examples of metadiscourse. This process was important to confirm my own initial interpretations, and led to the identification of additional interpersonal functions, functions that had not been evident during the first analysis, but it also led to the rejection of some functions for which there was then no support but my own intuition.

RESULTS AND DISCUSSION

Overall frequency of metadiscourse in sermon

As shown in Table 4 , metadiscourse is used extensively by preachers from all three denominations, reinforcing the received view that contemporary sermonic discourse is highly interpersonal. It is notable that there is very limited variation across the three denominations in terms of the distribution of metadiscourse (chi-square analysis supports this finding; Table 4 ). A total of 16,500 cases of metadiscourse were found in the collected sample (189,239 words).

Table 4:

Interactive and interactional metadiscourse in the sermon samples. Numbers indicate occurrences/1,000 words

	Anglican	Per cent	Baptist	Per cent	Catholic	Per cent	All	Per cent
Transitions	10.4	50	10.0	50	8.7	50	10.1	50
Frame markers	2.5	12	2.9	15	1.8	10	2.6	13
Endophorics	0.1	0	0.1	1	0.04	0	0.9	0
Evidentials	4.2	27	5.1	25	5.0	29	5.1	27
Code glosses	2.1	10	1.9	9	1.9	11	2.1	10
Interactive	19.3	23	20.0	23	17.4	21	20.6	22
Hedges	12.1	17	11.5	17	7.7	12	11.5	16
Boosters	8.0	11	6.9	10	6.8	10	7.5	11
Attitude markers	2.1	3	1.9	3	1.9	3	2.0	3
Engagement markers	37.7	53	37.9	56	38.9	60	39.5	56
Self-mentions	11.2	16	9.4	14	9.2	14	10.3	15
Interactional	71.1	77	67.6	77	64.5	79	70.5	78
Total metadiscourse	90.4		87.6		81.9		91.1

Note : Chi-square analysis of denominational differences pertaining between categories of interactive metadiscourse was marginally significant: χ ² = 14.40, df = 8, p = 0.07.

Chi-square analysis of denominational differences pertaining between categories of interactional metadiscourse was significant: χ ² = 64.72, df = 8, p < 0.01.

Chi-square analysis of denominational differences at the overall level of interactive metadiscourse (all categories combined) and interactional metadiscourse (all categories combined) was not significant: χ ² = 3.92, df = 2, p = 0.14.

For every 1,000 words in a sermon, between 8 and 9 percent, or approximately 80–90 words, are words (or phrases) that are used with a metadiscursive function. Assuming that preachers speak at a normal speaking rate of 160 words per minute, they use a metadiscursive resource of some kind once every 4.5 s during a 7.5-min sermon.

The only available ‘norm’ against which these findings can be compared is Boggel’s (2009) diachronic study of metadiscourse in sermons from the Middle English period (13th century) to the Early Modern English period (18th century). Boggel recorded approximately 10 tokens of metadiscourse per 1,000 words of 18th century preaching—the reported total frequencies for metadiscourse in contemporary sermons are almost 10 times higher than that. This suggests that, over time, a significant interpersonal shift has taken place, with contemporary preachers being far more likely to deploy metadiscursive resources to reach out to sermon listeners. The shift is to be expected because the contemporary preaching event is subject to fundamentally different postmodern contextual conditions.

The overall occurrence of metadiscourse reported in Table 4 is also considerably higher than that reported for some other contemporary, academic as well as non-academic genres,for example, postgraduate dissertations ( Hyland and Tse 2004 ), research articles, academic textbooks, CEOs’ letters, and Directors’ reports ( Hyland 2005 ). However, closer inspection of the frequencies for individual categories of metadiscourse reveals that the higher overall frequency in sermonic discourse is attributable to substantially higher frequencies of boosters, self-mentions, to some extent hedges, but particularly to many times more engagement markers, suggesting that the spoken religious context governing the use of metadiscourse during preaching calls for very different metadiscursive practices.

Patterns of metadiscourse in preaching

In the sermon samples, across all three denominations, preachers use more than three times as much interactional metadiscourse (70.5 occurrences/1,000 words in the combined sample) as they use interactive metadiscourse (20.6 occurrences/1,000 words), indicating that preachers are conscious of involving listeners in the sermon content, but less concerned (relatively speaking) with guiding listeners through the sermon structure.

Recent studies ( Cavalieri 2011 ; Ädel 2012 ) have indicated that spoken discourse may indeed be more dependent on interactional than interactive metadiscourse because of the immediacy of face-to-face communication and other (oral) conditions imposing on the communication, but those reasons, by themselves, are not enough to explain the very large difference between interactional and interactive metadiscourse accounted for in this type of (written-to-be) spoken communication. Part of the answer may instead be attributable to the fact that preaching essentially assumes and ‘builds on an ecclesial relationship’ ( van Seters 1991 : 269) thus, that preaching is an ecclesial act (and therefore different from a similar act involving someone speaking and an audience in a secular setting). Regardless of whether the basic purpose of contemporary preaching is kerygmatic (emphasizing the spirit of the Gospel), transformational (facilitating an experience of God), or conversational (sharing the responsibility of scripture interpretation) ( Rose 1997 ), it is ultimately a move away from the purpose of ‘traditional’ preaching (taking a text and applying it persuasively); although preaching is still nourished by scripture, the ‘text’ is no longer central, meaning that an understanding of the sermon as text also becomes flawed. In order to move listeners to believe, listeners must ultimately, van Seters ( 1991 : 270) says, be made to ‘know and feel what this [belief/faith] means’. It is possible those interactional metadiscursive resources serve this New Homiletic-performative purpose of the sermon to a better end and that they are therefore more frequently employed by preachers in trying to achieve their preaching objectives.

Turning our attention to the most important sub-categories of interactional and interactive metadiscourse, it is notable that the most frequent category of all is engagement markers (39.5 occurrences/1,000 words), comprising 43 percent of all metadiscourse combined, reflecting how fundamentally important the explicit preacher–listener relationship building is in contemporary preaching. Thirteen per cent of the metadiscourse is made up of hedges (11.5 occurrences/1,000 words), making it the second most important sub-category, and highlighting how vital it is for preachers to allow listeners space to individually explore and shape their faith. Of the interactive sub-categories, the presence of transitions (10.1 occurrences/1,000 words) and evidentials (5.1 occurrences/1,000 words) indicates that preachers are concerned about listeners’ coherent ideational navigation of the sermon, reasserting preaching as part of a hermeneutic tradition, and contradicting any claim that the sermon is monologic. The next section considers in more detail how these and other metadiscursive resources may be productive to meet expectations from New Homiletic preaching.

Contribution of interactional and interactive metadiscourse to preaching turned to the listener

Preaching Principle 1—Elevating listeners’ role

The principal way for preachers to address Principle 1 is by relying on engagement marking, whereby they can: (i) locate/name listeners in the sermon; (ii) acknowledge listeners as dialogic partners; and (iii) emphasize inclusiveness and solidarity, all of which are metadiscursive functions that ultimately serve to elevate listeners in the sermonic event.

When preachers name listeners directly, by recourse to the second person pronoun you , as in (5), they simultaneously place listeners in the room and enable listeners to ‘participate in the weaving into the fabric of the sermon’ ( van Seters 1991 : 277).

• (5) You are not obliterated; you have been given the Holy Spirit, the Lord and Giver of life.

Preachers are apparently sensitized to the implications of this kind of listener engagement, as indicated by this comment from an Anglican preacher:

You have to be direct in the way you address your audience. I tend to use “you” a lot, simply to acknowledge them, their needs, their desires, and what have you.

While naming listeners in this way is a most direct way of bringing the listener into the sermon, it also brings with it the unfortunate implication of either including or excluding the preacher, something most preachers want to avoid, as noted by this Catholic preacher:

Sometimes I catch myself saying “you” and I’d then say: “Of course I include myself in that”.

Using engaging questions like that in (6) during preaching is a way for preachers to encourage listeners to be active and responsive during listening, thereby promoting them from being passive recipients of ideas to conversational partners in a construed dialog. A Baptist preacher commented on the listener-elevating effect of questions in what is widely believed to be a monologic discourse this way:

During a sermon the listeners are rendered rather passive but what you try to do is you try to enter into conversation with them. Questions are absolutely crucial in this conversation because they turn your monologue into a dialogue of sorts. Questions are an instrument for showing listeners that the sermon is about them, they are included, involved.

• (6) What is your darkness?

Questions, open or closed, genuine or rhetorical, are effective engagement strategies also because preachers can use them to alternate between a ‘teaching’ position (where they are empowered to ask questions) and a more common ‘student-of-faith’ position, where preachers stress their desire to join the listeners as questioning hearers of the word ( Immink 2004 ), very much in search of answers rather than in a position to provide them, as expressed by this Anglican:

I ask myself the very same questions of faith as the people in my congregation do, and I do not mind showing them that.

Most commonly, however, preachers appear to use engagement markers to accentuate inclusiveness and solidarity with listeners, to construe preaching as a ‘shared activity of preacher and listeners’ ( Waznak 1998 : 24), thus elevating listeners by emphasizing preacher–listener equality. This is manifested by very extensive use of different forms of inclusive first person plural pronouns, that is, ‘we’, ‘us’, ‘our’, ‘ours’. Stressing inclusiveness is a function of the preacher’s overall ambition to communicate that preacher and listener make up one body, ‘the body called “church”’ ( van Seters 1991 : 269), were both parties are guided by the same confessional rationale, and sharing a common set of beliefs, as shown in example (7).

• (7) We believe that life is gift, not granted us for free.

Preachers are conscious of the implications of communicating this inclusiveness through their preaching, as we can see from some of the comments from the interviews:

I very much use “we”, “we” and “us”. “We” as a people, “We” as a church. [Catholic]

I’d say we are very conscious of inclusive “we”. It is very popular in the Church of England at the moment. There is a bit of pressure for that. [Anglican]

Preaching Principle 2—Opening up an interpretative space for listeners

According to Wesley Allen ( 2010 : 8–9), sermon listeners must be granted the ‘freedom to assent or disagree’ with a sermon. This freedom presupposes that there is an interpretative space in the sermon that is open to the extent that listeners may inhabit it and exploit it, informed by their experiences, opinions, and desires. The principal way for preachers to achieve this metadiscursively is by recourse to hedges. According to Hyland hedging reflects a desire not ‘to express commitment [to the truth value of an accompanying proposition] categorically’ ( Hyland 1998 : 1). To this end, hedges are used metadiscursively to: (i) encourage listeners to offer alternative interpretations; (ii) invite listeners to explore further on their own; and (iii) to recognize doubt among listeners.

In (8) it is clear that the preacher offers his own interpretation of why Peter was so appalled by Jesus’ suggestion that he should wash Peter’s feet—the place in the Bible that this refers to does not tell us why. In keeping with Principle 2, the preacher thus employs the hedge ‘perhaps’ to signal that this is not a fact-based claim, simultaneously inviting the listeners to provide their own explanation for Peter’s reaction.

• (8) Why was Peter so appalled? Perhaps Peter’s worldly wisdom had trained him to be very suspicious about gestures of service.

When we discussed this extract during the interview, the Baptist preacher confirmed this function of hedges and justified his use of ‘perhaps’ and other hedges like it thus:

As soon as I offer an interpretation of some place in scripture, I need to let people know that this is just my take on it, my interpretation. I think words like “perhaps”, “maybe”, “possible” and the like are a kind of door opener to people listening, telling them to offer their view on the reading.

Interestingly, when our discussion later centered on another instance of hedging, the same preacher highlighted that hedges also:

[…] encourage people to look at it for themselves. [Hedges are] a way for us to encourage listeners’ individual exploration of the Bible.

Both these metadiscursive functions of hedging (prompting listeners to supply their own alternative interpretation and inviting listeners to explore further) are consistent with homiletic advice that ‘a sermon must be open to many interpretations and hearers are invited to complete the sermon’ ( Immink 2004 : 100) and thus directly address Principle 2.

In example (9) the hedge ‘may’ acts ‘catalytically’ upon a sermon theme of ‘doubt’ and serves to acknowledge listeners’ religious uncertainty.

• (9) We may have lost the old certainties that we could hold when we weren’t challenged every day by a diverse sea of faiths and of moral ambiguity.

When asked to comment on his use of ‘may’ in this excerpt the preacher expressed the view that:

You have to acknowledge that there are huge amounts of doubt around and the people in Church are not immune from this and nor is the preacher himself. […] By being cautious in this way I can acknowledge the listeners’ feeling of uncertainty. [Catholic]

Preaching Principle 3—Avoiding a voice of authority, yet confirming faith

In a recent book, Wuthnow ( 2012 : 38) notes the ‘absence of any empirically verifiable knowledge’ in religious contexts and says that ‘the existence of God cannot be proven; nor can many other claims about God (…)’. In contrast to secular discourses, many claims made in sermonic discourse (and other religious discourses) typically resist a characterization as true or not true; this opens up an interesting epistemological continuum which must be addressed communicatively in preaching turned to the listener. Although most preachers do not want to be seen as dogmatic, an excessively tentative ‘anything goes’ attitude is unlikely to resonate well with listeners (notwithstanding Principle 2). In order to avoid a voice of authority but simultaneously allowing themselves to confirm the faith that is being preached, that is, to address Principle 3, preachers may engage in a metadiscursive act of balance involving hedges, boosters, and evidentials. Preachers make frequent use of hedges to (i) balance the authority inherent in the sermonic institution; and to (ii) assume a humble attitude and to recognize their failure to understand something completely. As an epistemic pendant, boosters (7.5 occurrences/1,000 words) may be used to reconfirm shared convictions held between preachers and listeners. Finally, preachers can make use of evidentials and rely on the resulting multivoicedness to tone down their own authority indirectly by momentarily giving up the sermonic floor to other sources.

Buttrick ( 1987 : 249) notes that ‘preaching is usually understood in relation to authority’. Thus, if authority is an undesirable given, preachers will want to ‘balance’ that authority in keeping with Principle 3. A balancing function attributed to hedges was apparent when I discussed example (10) with the Roman-Catholic preacher who is the source of the sermon.

To some extent, I am an authority because I represent the Church but I need to be careful with how I negotiate and trade on that authority. Like Jesus, I want to lead people to God, but his words and my words address people in completely different historical contexts. Where Jesus could command people, I definitely need to be more tentative with how I direct my listeners.

• (10) Perhaps you can do whatever little is in your power to change the lives of these people.

The preacher’s words echo the proviso from Wesley Allen ( 2010 : 8) that contemporary listeners ‘must be able to “apply” the word spoken to their own lives in their own ways without it being dictated or there is no freedom’.

A toning down of authority is apparent also on those occasions when hedges seem to be used by preachers to acknowledge that they do not have the ultimate answer or all the relevant facts, as with ‘might’ in (11), where the hedging function is prefaced by an explicit statement to that effect and therefore easily identified.

• (11) It is difficult to understand what Paul is getting at. Might he be implying that turning away from sin is the answer?

When preachers use boosting language, their rationale is neither to express a voice of authority, nor to close down dialog per se (the stereotypical functions of boosters). Instead preachers adopt a voice of assuredness ( Reid 2006 ) to emphasize epistemic allegiance with listeners, making the discursive bond pertaining between preachers and listeners stronger. Propositions metadiscursively framed by boosters are either objectively grounded (based on historical facts or empirical evidence), or subjectively grounded (based on/in faith or conviction, and without empirical support). Examples (12) and (13), both containing the boosting verb ‘know’, illustrate an objectively and a subjectively grounded proposition respectively.

• (12) We know that the early Church, after the resurrection, built itself up around those celebrations.

• (13) I know that the resurrection is not just about my personal relationship with God.

When interviewed about his choice to use ‘know’ in (12) the Anglican preacher expressed the view that:

This is common knowledge, something we know for certain […] I can be firm because there are so many different sources that establish this for us, so there is really no need to debate this .

The Baptist preacher who is the source of (13) responded to my question concerning what basis there is for the ‘know’-assertion by saying that:

There are some aspects of our faith where we do not need empirical evidence to express a fundamental conviction that something is true. The Holy Spirit sometimes provides all the evidence we need. If I did not express myself with assuredness regarding something as basic as our resurrection I might as well take a vow of silence. My listeners expect this assuredness from me when I talk about something we together can treat as given.

Authority may also be manifested as vocal discourse dominance, and by virtue of the apparently monologic nature of sermons, preachers are in danger of being seen as dominant/authoritative in this respect. However, by allowing other voices to enter the sermon, the dominance is, momentarily at least, disrupted; for this reason, the use of evidentials, as in (14) may be seen as preachers’ reactions to their own vocal dominance, their right to be the sole voice on whatever topic is at hand, and consequently as playing to Principle 3.

• (14) C.S. Lewis once said : ‘Pain is God's megaphone’.

Preaching Principle 4—A sermon design that helps listeners to engage with the sermon during preaching

Buttrick ( 1987 : 70) talks of sermons as ‘a sequence of ideas, moves, within a movement of thought’, and ideas in sermons should be connected by ‘logical coherence’ lest they ‘may enter the consciousness as baffling non-sequitur’ (the latter would be diametrically opposed to Principle 4). This assumes that some sort of structure is adhered to by preachers in preaching turned to the listener ( Lowry 2001 ; Davis 2003 ; Long 2005 ). To address Principle 4, preachers may use transitions to provide listeners (i) an opportunity to ‘catch up’ and ‘get back on track’; (ii) reassurance regarding understanding; and (iii) an opportunity to re-evaluate their understanding of the sermon. Furthermore, preachers can employ frame markers (2.6 occurrences/1,000 words) to: (i) articulate that a sermon is made up of sections and stages; (ii) signal shifts in focus; and (iii) introduce a momentary and effective pause in the flow of the sermon. Finally, code glosses (2.1 occurrences/1,000 words) can be used to secure listeners’ understanding.

Even if turn-to-the-listener preaching is typically ‘less linear’ than traditional preaching ( Long 2005 : 170), coherence of discourse is a central concern. This would explain the fairly common occurrence of transitions, like contrastive ‘however’ in (15), the role of which is to help listeners to steer their way through the sermon.

• (15) It is tempting to call the women disturbed, and they may well have been. However , Jesus did have an affinity with the marginalized.

Practicing preachers acknowledge both the need for transitions in their sermons and the fact that transition markers perform different metadiscourse functions; this is what an Anglican priest said when asked about what measures he takes to make sure the congregation can follow the evolving sermon.

When I signal how I would like us to move from one aspect of the sermon to the next I do that not only to help them make sense, but also to afford everybody a chance to catch up with what we are talking about.

Another function of transition markers is thus to act as discursive ‘checkpoints’, allowing listeners to (re-) evaluate their understanding of the sermon content. Long ( 2005 : 148–9) explains this function of sermon transitions as providing ‘reassurance [to listeners] that they have, indeed, gotten the message. […] For those hearers who have become confused, however, or whose minds have wandered, the [transition] provides an opportunity for them to re-enter the sermon flow, to get back on track’ ( Long 2005 : 148–9). Prompted by our discussion regarding his use of ‘therefore’, a Baptist preacher confirmed these functions of transitions making the point that:

Words like “however”, “therefore” etc., although they are quite academic in nature, provide a sense of encouragement to my audience to follow what I say. Interestingly, words like these also help people return to the sermon, e.g. if they are distracted, which sometimes happens.

The basic rationale for transitions, to add to sermonic coherence in keeping with Principle 4, is the same for frame markers. Notwithstanding advice from homiletics to limit the use of framing expressions because they introduce an element of ‘time-consciousness’ in the sermon, risking increasing ‘congregational restlessness’ and creating a ‘tension between enumeration and content’ ( Buttrick 1987 : 70) several of my informants, like this Catholic priest, were adamant that articulating that the sermon is made up of sections and stages, as indicated by the frame markers in (16), is important for listeners.

Whenever I preach I try to express quite explicitly to the congregation where I am in the sermon. I think this is important so that they can understand how different parts fit together.

• (16) First , we can embrace it […]. Second , we can call on […].

The analysis suggests that preachers use frame markers also to signal important shifts in focus, as in (17) where the preacher uses ‘so’ effectively as a demarcation between the preceding section of the sermon (an explanation of a place in the Bible), and the section that follows after the rhetorical question, where the focus is transferred from the text to the listeners.

• (17) So , my brothers and sisters, where are you this morning?

For prosodic reasons, frame markers usually introduce a momentary pause in the flow of the sermon during which listeners are allowed the opportunity to re-view the ongoing discourse as a sequenced narrative, and to revisit the assumptions held to that point in the discourse about the sermon content and the opportunity to change them. This affords sermon listeners a chance to understand the sermon content in a potentially clearer way when the contents is not presented as a continuous flow of ideas based on the same assumptions.

Nearly all preaching has an expositional component to it ( Lloyd-Jones 2012 ), but exposition/elaboration is sometimes metadiscursively marked in the sermon by recourse to code glosses, such as, ‘that is (to say)’, ‘for example’, or ‘namely’ as in (18), helping listeners engage with the sermon during preaching by signposting portions of the sermons that elaborate contents in some way.

• (18) Jesus acts as a hired hand, as one without status, namely , as a servant.

Preaching Principle 5—Focusing on preaching as embodiment and performance

Turn-to-the-listener preaching is firmly grounded in embodied ‘experience’ and performance ( Craddock 2001 ). Sermonic meaning is created performatively in the very act of preaching, meaning that how something is said becomes vitally important, and listener-oriented preaching must be infused with emotion and passion – Wesley Allen (2010) talks of New Homiletic preaching being governed more by pathos than by logos. The chief means for introducing emotion and passion linguistically (disregarding body language), that is for addressing Principle 5 metadiscursively, is attitude markers (2 occurrences/1,000 words), like ‘wonderful’ and ‘good’ in (19).

• (19) That is the wonderfulgood news of our faith.

When expressing attitudes, preachers appear to favor positive attitudes, expressed by words like ‘important’, ‘amazing’, ‘interesting’, and ‘good’ – indeed, preachers are more than five times more like likely to qualify something as ‘good’ than as ‘bad’ – thus taking their responsibility to be ‘heralds’ of the ‘good news’ ( Waznak 1998 ) very seriously.

Preaching Principle 6—Preaching is inductively shaped, as a coherent narrative on the basis of lived experience about faith

An influential form(ula) applied in various shapes to New Homiletic preaching is to have sermons ‘move from the specifics of lived experience to general claims’ ( Wesley Allen 2010 : 8–9); an inductive retracing of the preacher’s experience of how general meaning/understanding may be arrived at. This kind of retelling-of-experience approach often assumes a different structure compared with traditional, deductive, and argumentative preaching shaped on the basis of a ‘three points-two-jokes-and-a-poem’-design, and instead a turn-to-the-listener sermon commonly assumes various forms of a narrative structure ( Wesley Allen 2010 ). An important means for preachers to address Principle 6 metadiscursively is with reference to self-mentions (10.3 occurrences/1,000 words) and evidentials. Self-mentions: (i) allow preachers to draw on lived experience; (ii) add authenticity to the narrative; and (iii) add a ‘witness’ perspective. The witness dimension is evoked also by the use of evidentials that enable preachers to introduce testimonials of other people. Transitions and frame markers provide narrative coherence as explained under Principle 4 above.

The sermon excerpt in (20) highlights the powerful effect of drawing on lived experience. The preacher uses a story from his own life, describing how, while he was living his vocational dream, all of a sudden he was diagnosed with a serious disease. The theme for the sermon is ‘Jesus knew sorrow as well as joy’ and by using an example from his own life, in the form of a story of joy and sorrow, the preacher hopes to bring the essence of the sermon theme across to the listeners in a better way; the fact that the preacher himself is the main protagonist in the story adds to the authentic nature and closeness of the narrative and, at the same time, the preacher offers a testimony of the very faith he is preaching.

• (20) I was ordained a deacon in June 2006, and I entered ministry full of joy. […] three years of training in theological college, I was finally doing that work. When, in August 2006, my [describing a medical condition] and I went to the doctors. I was devastated. […] We can trust God with our sorrows as well as our joys – it helped me .

The metadiscursive functions of self-mentions, and their relation to Principle 6, came out clearly during the interviews when self-disclosure was discussed.

I include myself because I am a witness of faith.

Your own experience always carries a greater degree of authenticity in relation to the point you make. I draw on my own experience very consciously. […] The story sermon is now in many ways the established form for preaching, for good reasons. I sometimes use myself in the story, and sometimes others, but in all cases it is vital that preaching is about [emphatic] someone, not just about an abstraction.

When someone other than the preacher offers testimony/features as a witness/is the source of the lived experience referred to, preachers can rely on evidentials to project that person into the sermon, as in example (21) where the preacher relates a story from a journey during which he visited a charismatic church service in the US.

• (21) Steven was unstoppable. He talked without interruption for at least 30 min of his ‘joy at seeing the resurrected Christ’ in the eyes of this homeless man.

CONCLUSIONS

This investigation of data samples of preaching in three well-established preaching traditions in England suggests that metadiscourse holds an absolutely central place in contemporary preaching. Setting out, the study established three research foci, and finding that: (i) metadiscourse is prolific in preaching across denominational boundaries; (ii) preachers appear to favor interactional metadiscourse (particularly engagement markers) over interactive metadiscourse; and (iii) preachers’ metadiscursive practices can be productive to address goals and expectations characteristic of preaching turned to the listener, raises interesting implications of this research.

Most importantly, the conclusions of this study can be applied to the field of homiletics, in two ways. First, they provide a description of effective metadiscursive use. Much in the same way that descriptions of conventional features of academic discourse has enabled the teaching of English for academic purposes (see, e.g. Hyland 1998 ; Hyland and Tse 2004 ), the descriptions emerging from this research can sustain approaches to teaching homiletics to pre-service and in-service practitioners. If preachers are made aware of the powerful interpersonal means provided by metadiscourse they can evaluate their own as well as others’ preaching in a new light and consequently make informed choices regarding language use and turn preaching more appropriately to the listener. Secondly, although the sermon is an expressive vehicle in which the form of language plays a critical role in achieving the desired effects, advice from within homiletics concerning language use is typically prescriptive and only very rarely informed by empirical research. It is hoped that this investigation of the place of metadiscourse in sermons will add descriptive nuance to discussion in the wake of the New Homiletic about how preachers can explore language and engage listeners in congregation-centered hermeneutics. This discussion could use the concept of metadiscourse to explain how turn-to-the-listener preaching is as much a matter of acknowledging and negotiating social relations with listeners through language as it is a means of using language to move listeners to believe.

A broader implication of this work is that it adds to the small number of linguistic investigations that have applied a model of metadiscourse to non-academic discourse. There are indications that the prominent and important role of metadiscourse is not just a feature of academic writing but rather a ‘coherent and principled means’ ( Hyland 2005 : 141) for explaining communicative interaction in various kinds of discourse. However, the research to date has been so very heavily disposed in the direction of academic discourse that little evidence for this broader role exists (but see Dafouz-Milne 2008 ; Jensen 2009 ; Cavalieri 2011 ). This article constitutes an additional contribution to this small but much needed ( Ädel 2010 ) body of data from other domains.

More generally, this work serves to demonstrate the value of using established frameworks of analysis in other linguistic domains. The heavy concentration on academic discourse has allowed the research in applied linguistics to acquire a significant depth. By extending investigations to other domains (like religious discourse and, more specifically to the genre of sermons), we can not only confirm whether findings apply to other areas of language use, we can confirm (or disconfirm) the utility of analytical approaches and frameworks, thus enhancing the ability of researchers in the field to apply the findings of linguistics research to benefit a broader range of consumers.

Conflict of interest statement . None declared.

NOTE

1 This principle has been counteracted by Ifantidou (2005) .

ACKNOWLEDGEMENTS

The research for this study is supported by the Swedish National Bank (Riksbankens Jubileumsfond). The author also wishes to thank two anonymous reviewers, the Editor, and Professor Diane Pecorari for constructive comments on previous versions of the manuscript.