Chapter 1 Introduction

Ten years ago, I remarked on the surprisingly small proportion of metadiscourse studies that investigated spoken discourse compared to the total amount of metadiscourse research (Mauranen 2012). This is still true, even though much more research has been devoted to spoken discourses since then. The proportional gap remains enormous. If we look for studies addressing not only speaking in general, but speaking in dialogic interaction, the result is hardly visible to the naked eye.

The ultimate reason for studying metadiscourse has not changed. It is the intrinsic fascination of this fundamental characteristic of human language: the ability to reflect on itself. This inbuilt capacity in our languages is an indication of a more general capacity of the human mind to monitor its own operations, that is, metacognition. We can think about our own thinking, and we can talk about our own talking. Language is nevertheless not only an instrument of cogitation, but of communication. Dialogic speech is where cognition meets interaction; metadiscourse is one of the resources that language has for increasing the transparency of our intended meanings and communicative intentions to our interlocutors. It builds on our theory of mind, which makes assumptions about what our interlocutors or audiences know and think, and thereby helps us design our talk for our recipients accordingly. Importantly, metadiscourse is a discourse phenomenon, therefore not reducible to individual words or phrases, and even when individual metadiscoursal expressions coincide with a word or phrase, it is their status in the discourse that matters. Therefore, counting ‘metadiscourse markers’, useful as it may be for comparisons, generally makes for conservative rather than innovative research.

Speech is foundational to language, unquestionably its most ubiquitous and constant mode of use, and can with a high degree of confidence be said to be its original mode (possibly vying for first place with sign language); spoken interaction is what language is fundamentally about. Passing it over in studies of metadiscourse is a major omission.

Two things, then, motivate writing a book on metadiscourse in spoken interaction: metadiscourse research has all but ignored spoken interaction, and spoken interaction research has all but ignored metadiscourse.

What reason do we have, then, for assuming that investigating dialogic speech might bring new understanding to the study of metadiscourse? Most studies comparing written and spoken metadiscourse, or only studying the latter, have found no major differences. The early studies that compared metadiscourse in speech and writing discovered only minor differences (Luukka 1995; Ädel 2010), and although some more recent studies have begun to challenge their similarity somewhat more (Lee & Subtirelu 2015; Liu 2021), they have not come up with radical departures either. Even without direct comparison, studies of spoken academic monologues like lectures or presentations have applied analytical models built on written texts and found largely similar metadiscourse (e.g., Rowley-Jolivet & Carter-Thomas 2005; Webber 2005; Pérez-Llantada 2006; Fernández Polo 2018), with some scholars observing more colloquial expressions (Flowerdew and Tauroza 1995; Zareva 2011). More recently, these more traditional academic speech genres have received an addition from a short presentation type, the three-minute pitch, or three-minute thesis presentation (3MT), where doctoral students present their research in competitive settings. The 3MT has become a popular topic for metadiscourse research (Zou & Hyland 2020; Hyland & Zou 2022; Qiu & Jiang 2021; Liu 2021) and other kinds of discourse studies. Again, the studies have applied Hyland’s (2005) writing-based model with no major alterations or additions, and again the mode is monologic.

New light is thrown on the question in Zhang’s (2022) recent large-scale multidimensional study on reflexive metadiscourse across spoken and written registers. The study includes dialogic, that is, conversational registers, and unlike the monologue-focused earlier studies, finds a major divide between speech and writing. Zhang shows that spoken registers, unlike written, use metadiscourse markers to explicitly emphasize interaction between the addresser and addressee. That is, conversational speech typically displays frequent occurrences of metadiscourse markers relating to addresser, addressee, saying and arguing, receiving and understanding, in addition to defining and explaining, which it shares with other registers. The study thus reveals significant differences between dialogic and monologic registers: what Zhang calls markers of ‘participant interaction’ are frequent in dialogues but not in monologues. At the same time, what he calls markers of ‘discourse presentation’ show no significant difference between dialogues and monologues. The cross-modal similarity found in previous studies may thus to some extent be explained by characteristics of monologues.

While dialogic speech has remained a rare option in metadiscourse research, it has not been entirely non-existent. In addition to Zhang’s (2022) inclusive comparison, McKeown & Ladegaard (2020) studied metadiscourse and dominance in multi-party discussions, and I have also carried out a few smaller-scale investigations (Mauranen 2001, 2003, 2010, 2012). To my knowledge there is even less on written dialogic metadiscourse, with the exception of three studies of digital dialogues, Smart (2016), Biri (2021), and Mauranen (2021b).

Given that there is so little research on spoken dialogue – including polylogues – the question of how it might differ from research based on monologic discourse, whether spoken or written, is motivated on that basis alone. Dialogue is synchronous joint activity, which can be symmetric, that, is, participants can perform similar actions. By contrast, monologic speaking is asynchronous, performed by a single speaker, the activity thus neither joint nor symmetric. Many intriguing questions arise regarding the effect that the joint, synchronous, and symmetric activity, the simultaneous presence of interlocutors and the rapid pace of turn-taking in dialogue might have on metadiscourse.

Interacting in real time means that participants in a dialogic exchange need to share their representations of the situation to a sufficient degree to be able to carry out a meaningful dialogue, that is, they must ‘match their perspectives’, or, in Pickering and Garrod’s (2021) terms, they must be aligned. Does metadiscourse play a role in this? If there is little time for pre-planning, and a pressing need to attend to what interlocutors are saying to make the speaker’s own speech relevant to how the interaction is going, or to steer it to their preferred directions, how, if at all, might metadiscourse come into it? Will new functions or uses emerge, beyond possibly some new expressions or well-known features of speaking, like hesitations or colloquialisms? How do monologues, spoken dialogues, and written dialogues compare? Spoken monologues and dialogues are both embodied and fast unlike written dialogues, while all dialogues have at least two active participants, which implies that the outcomes and directions of the discourse are far less predictable than in monologues of any kind. Can we find social or other situational parameters that affect metadiscourse in interaction?

Clearly, the distinct character of metadiscourse in the conversational register compared to other registers (Zhang 2022) adds to the interest value of discovering what could be behind the finding on closer inspection.

Previous research into spoken metadiscourse has not only been heavily biased towards monologic genres, but also favoured academic discourses. There are some exceptions to the academic emphasis that I am aware of, like Ilie (2003) on parliamentary debates and Malmström (2014) on preaching, and hopefully there are more to come. Studies on metadiscourse in the digital media have shown signs of opening towards other domains, like notice boards (Smart 2016), social media (Biri 2021), advertisements (Delibekovic Dzanic and Berberovic 2021), and vlogs (Ädel 2021). The academic bias is nevertheless true of metadiscourse research overall, perhaps reflecting its roots in student writing in Crismore’s (1983) and VandeKopple’s (1985) work. All the studies mentioned above on monologues like lectures, conference talks, and 3MTs have taken their data from the academic domain.

In this book, I intend to tread the same path. This may at first sight look like a strangely limiting choice: surely ordinary conversations are more fundamental to dialogic speaking than academic discussions? There are nevertheless two good reasons for doing this. The first is that in extending a well-charted research area like metadiscourse studies towards new territory, it makes sense to keep some basic parameters intact to be able to relate the results to previous findings. This means that any new insights from the dialogue focus can be seen against the background of what we already know about metadiscourse. The second reason is that academia overall is highly reliant on not only writing but on the spoken word. Although in many experimental and laboratory sciences, the core research may not appear to depend on...