Vi snakker sådan: An analysis of the Danish discourse-pragmatic marker sådan

Abstract

The last three decades have witnessed increasing interest in discourse-pragmatic markers (DPMs), both with regards to their high frequency in spoken discourse and their multifunctionality in interaction. Most studies have centered on English, with studies on Danish restricted to a handful of previous interactional discourse analyses. This paper is a preliminary investigation of the Danish word sådan (commonly glossed as ‘such’ or ‘like this/that’). A qualitative, form-based, discourse analytic approach is undertaken on over 40 minutes of naturally occurring Danish talk to argue that sådan qualifies as a DPM. In service of textual, subjective, and intersubjective macro-functions, sådan illustrates; exemplifies; marks hesitation; approximates a quantity; mitigates, hedges, or softens; and allows self-correction or self-repair. These findings are discussed in terms of their implications for sådan’s place in the Danish DPM system and our understanding of DPMs across languages.

1. Introduction

The last three decades have witnessed burgeoning interest in discourse-pragmatic markers (henceforth DPM in singular and DPMs in plural). Examples of these syntactically optional, semantically ‘shallow’ words or phrases include: like, I mean, well, and you know in English (Beeching 2016); ja, doch and wohl in German (Gast 2008); and da, jo, nu and vel in Danish (Davidsen-Nielsen 1996). Research into DPMs began when scholars recognized their importance in ‘carry[ing] social meaning, perform[ing] indispensable functions in social interaction, and constitut[ing] essential elements of sentence grammar’ (Pichler 2010:582). DPM studies have taken a wide swath of theoretical and empirical angles, including pragmatics (Fraser 2006), discourse (Schiffrin 1987), conversation analysis (Hakulinen 1998, Heritage & Sorjonen 2018), language attitudes (Dailey-O’Cain 2000), computational linguistics (Popescu-Belis & Zufferey 2011), Relevance theory (Andersen 2001, Vivien 2006), and sociolinguistic variation (Pichler 2013, Diskin 2017). As one might expect from such a diversity of approaches, there exists an abundance of labels and definitions for what we refer to here as DPMs (Maschler & Schiffrin 2015). This paper subscribes to Pichler (2013), who defines DPMs as (i) syntactically optional elements that (ii) do not contribute to truth conditionality and that (iii) can structure discourse, express speaker stance, and/or guide utterance interpretation. We examine the terminological landscape, position our paper, and expand on Pichler’s (2013) definition in Section 2.

Although DPM studies have analyzed languages as diverse as Bislama (Meyerhoff 2002) and Chinese (Ljungqvist 2010), most have focused on English (Jucker & Ziv 1998a:2; Overstreet 2005:1846; Aijmer & Simon-Vandenbergen 2011:232) albeit with several important early papers on German modal particles (Aijmer & Simon-Vandenbergen 2011:223). The over-representation of English DPM studies tracks with the rise in Anglo-American interest in taking a discourse-pragmatic perspective, as opposed to the grammarian perspective popular in Europe (Watts 1988). However, it is problematic to overly rely on one language to formulate discourse-pragmatic related frameworks. First, it leaves unclear to what extent lessons learned and generalizations made from English DPMs extend to other languages. Second, overreliance on English can cause DPM functions to be overlooked (Schourup 1999). Third, it might result in what Redeker (1991:1170) calls ‘biases due to language-specific peculiarities in lexicalization’. Given the putative indispensable communicative functions of DPMs, these are important concerns.

Discourse-pragmatic marker research on the Danish language comprises about one to two dozen studies focused on words and phrases including altså, du ved, nok and vel (see Section 3.2 below). However, 25 years after Andersen (1986) called Danish discourse particles sprogvidenskabens stedbørn [linguistics’ stepchildren], Hansen & Heltoft (2011:1109) acknowledged they were dårligt udforsket [poorly researched], even within the well-established grammarian tradition (Watts 1988, Jakobsen 1995, Engberg-Pedersen et al. 1996). The lexical item sådan (glossed as ‘such’ or ‘like this/that’) is no exception to this relative lack of attention. This is despite uses as in (1), which, similar to previously studied Danish particles (e.g. Andersen 1982, Andersen 1986), appear neither to be syntactically required nor impact the utterance’s propositional content.

(1) Men min kollega ##((name)) havde faktisk også- altså de boede i så’n et ret lille Nørrebro lejlighed egentligt altså så har de bare det ligner så’n en lille, (.) og der kunne være sådan en brandslange inden i og så kan man sådan åbne det og så folder der bare sådan (.) pusleplads ud.

‘But my colleague ##((name)) actually also had- I mean they lived in så’n a quite small Nørrebro apartment actually so then they’ve got what appears to be så’n a little, (.) and there could be sådan a firehose inside it and then you can sådan open it and then out comes just sådan (.) a changing table.’

(C1:34-39)

In fact, sådan is almost entirely absent from Danish DPM discussions. To the extent it is mentioned, it is often as a component within studies on general extenders or the quotative system. Although some studies exist that tackle the form sådan (Jensen 2013a, b; Pedersen 2014), to our knowledge there exists no holistic, discourse-pragmatic analysis of the form sådan and its breadth of functions as suggested by usage in (1) above. Our paper seeks to fill this gap in Danish DPM research by focusing on the Danish lexical item sådan in the conversational discourse of young Danes (24–28 years old). This exploratory research constitutes, to the best of our knowledge, the first dedicated treatment of sådan as a DPM and the first broad overview of its functions in interaction. Our paper thereby contributes to research on Danish DPMs and to cross-language examinations of DPMs and their functions more broadly.

The research questions that motivated this paper are: (i) Does Danish sådan qualify as a discourse-pragmatic marker? and (ii) If so, what are the discourse-pragmatic functions of sådan in interaction? To understand the function of sådan in interaction, the first author collected and transcribed 43 minutes of conversational Danish data from students at a Copenhagen university cafeteria. Taking a qualitative, form-based (i.e. focused on the form sådan), discourse analytic approach, we first analyzed sådan’s frequency and then coded for its function. Based on our analysis, we contend that Danish sådan is indeed a DPM. We found it serves important textual, subjective, and intersubjective macro-functions in interaction. Sådan illustrates; exemplifies; marks hesitation; approximates a quantity; mitigates, hedges, or softens; and allows self-correction or self-repair. What’s more, participants used it with extremely high frequency,¹ making the dearth of prior discourse-pragmatic investigation into sådan all the more striking. The findings allow for future analyses of the idiosyncrasies and commonalities of sådan both in the context of Danish DPMs and of DPMs in other languages.

The paper is organized as follows: In Section 2, we clarify the DPM terminological landscape and contextualize Pichler’s (2013) definition. In Section 3, we review the dictionary and academic literature’s treatment of sådan, establishing promise for a DPM study and a gap thereof. Section 4 explains our qualitative, discourse analytic methodological approach and describes data collection and analysis. We present our answers to both research questions in Section 5, highlighting the multifunctionality sådan exhibits as a DPM. In Section 6 we discuss the implications of our findings and propose directions for future research, including examination of positional distribution, sociolinguistic use, similar DPMs in other languages, and Danish DPMs that display possible functional overlap with sådan. Section 7 is the conclusion of the paper where we summarize our findings and contribution to the field of DPM research.

2. Terminology

The heterogeneity of labels, definitions, and theoretical approaches to DPMs (see e.g. Maschler & Schiffrin 2015) requires a clear delineation of how we use the term ‘DPM’ here. Relevant to our discussion is the historical European and Anglo-American divide in DPM terminology. Broadly, the European school approached analysis from a grammarian perspective and preferred terms like ‘particles’ or ‘adverbials’, whereas the Anglo-American research traditions approached analysis from a discourse perspective and preferred the term ‘markers’ (Watts 1988). We see this in the Danish literature, with ‘discourse particles’ (Lundskær-Nielsen & Holmes 2010:413), dialogiske partikler [dialogic particles] (Hansen & Heltoft 2011:1033–1034), dialogiske adverbialer [dialogic adverbials] (Mortensen 2012), dialogpartikler [dialogue particles] (Christensen & Jensen 2014), and modalpartikler [modal particles] (Andersen 1982) as a few notable examples. For the purposes of this paper, we treat ‘particles’ just mentioned as representing the same phenomena of study as DPMs.

The DPM definition we employ in this paper comes from Pichler (2013), whose extended definition covers the multifaceted aspects of DPMs:

[D]iscourse-pragmatic features constitute a formally heterogeneous category of syntactically optional elements which make little or no contribution to the truth-conditional meaning of their host units and—depending on their scope, linguistic co-text, as well as sequential, situational and cognitive context—perform one or more of the following macro-functions: to express speaker stance, to guide utterance interpretation and to structure discourse. (Pichler 2013:4)

Pichler’s (2013) definition is embedded in the context of Halliday (1979), who proposed three modes, or functions, of meaning which language can serve: ideational, textual, and interpersonal. Ideational (using linguistic resources to talk about something) is not applicable to DPMs because it is concerned with the propositional content of utterances (Brinton 1996:38), but textual and interpersonal are useful categories that we will use in this paper.

The textual mode operates to distinguish text (either written or oral) from a collection of unrelated sentences. Pichler’s (2013) ‘structure discourse’ macro-function is synonymous with this textual mode. Example (2) illustrates how a speaker uses the English DPM ‘I mean’ to link two sequences whereby the second sequence clarifies or elaborates upon the first.

(2) I just think voluntary work is good because (.) you get to actually know how life works I mean you’re helping people pay their bills by taking some pressure off of them

(from Beeching 2016:187)

The interpersonal mode refers both to the speaker’s subjective attitudes towards utterance contents and to management of the social exchange. Pichler’s (2013) definition distinguishes between the speaker-centered stance aspect (subjective) and the dyad-focused utterance interpretation aspect (which she has defined as intersubjective; H. Pichler, personal communication, 30 May 2018). Following Pichler, this paper describes subjective and intersubjective macro-functions, which together constitute the interpersonal (Figure 1). It is important to note that the textual, subjective, and intersubjective realms are not mutually exclusive. DPMs are notoriously multifunctional and can operate in both the textual and interpersonal realms, even simultaneously (Andersen 2001, Cheshire 2007, Maschler & Schiffrin 2015, Beeching 2016).

Figure 1. Visual diagram of textual and interpersonal modes of meaning for DPMs.

Pichler’s ‘express speaker stance’ macro-function maps to the subjective mode of meaning. It is understood here as the speaker positioning himself/herself relative to interlocutors and cuing positions in interaction towards the discussion topic (Du Bois 2007, Tracy & Robles 2013). Table 1 illustrates several examples of how DPMs can function subjectively.

Table 1. Subjective functions of DPMs from Andersen (2001:67).

Pichler’s ‘guide utterance interpretation’ macro-function maps to the intersubjective mode of meaning. It is understood here as the speaker providing communicative clues to narrow the ‘meaning potentials’ (Thomas 1995:22) of what is said. This could involve effecting ‘cooperation, sharing, or intimacy between speaker and hearer, including confirming shared assumptions, checking or expressing understanding, requesting confirmation, expressing deference, or saving face’ (Brinton 1996:38). Examples (3) and (4) illustrate intersubjective functions performed with ‘right’, ‘you know’, and ‘like’.

(3) Sue lives in one of these renovated homes, right. And a lot of them are, you know, people that have roots here.

(from Denis & Tagliamonte 2016:91)

(4) Could I like borrow your sweater?

(from Underhill 1988:241)

In (3), ‘right’ and ‘you know’ indicate a presumption of the interlocutor’s common ground and can also serve as attention-seeking or adherence-eliciting markers – in Brinton’s words ‘confirming shared assumptions’ and ‘effecting intimacy’. In (4), ‘like’ hedges the request and allows the speaker to distance herself from its potential rejection – in Brinton’s words ‘save face’.

3. Previous studies

3.1 Sådan in the dictionary

According to Den Danske Ordbog [The Danish Dictionary] (henceforth DDO), the Danish word sådan comes from the Middle Low German sodan. Sodan comprises so, meaning ‘thus’ or ‘so’ in English, and the past participle of don, which was the verb ‘to do’ (DDO, accessed 9 April 2018). In Danish–English dictionaries such as Gyldendals Røde Ordbøger [Gyldendals Red Dictionaries], sådan is frequently glossed as ‘such’ or ‘like this/that’.

DDO provides two principal definitions for sådan, listed below in the order in which DDO presents them. English translations of the definitions and examples are our own. Uses marked with an asterisk (*) are those DDO flags as typical of spoken language. The DDO examples come from a 40-million-word corpus of written and spoken discourse, with particular attention paid to incorporating spoken language and discourse from everyday life. The corpus’ source materials, which span the years 1983–1992, cover both public (e.g. newspapers) and private (e.g. private letters) materials including books, radio, advertisements, diaries, oral interviews, and more.

The DDO entry contains further relevant information. First, it notes sådan in spoken discourse can undergo phonological reduction such that it is pronounced as så’n [sådan]. Schiffrin (1987:328) and Brinton (1996:33) have identified phonological reduction as a characteristic of DPMs, though Beeching (2016:6) notes that this does not hold true across the board (e.g. it is not always the case for ‘I mean’ or ‘like’ in English).

Second, the number and breadth of functions listed under the DDO definitions indicate one of the hallmark characteristics of DPMs – multifunctionality – and illustrate how sådan depends on context to be fully understood. In (1a, b) above, sådan refers to something else in the discourse, either in the recent past or near future. In 1e, this role is elaborated, as sådan further stands in for a lengthier concept. Because these definitions refer to something else in the discourse, they would fail to meet the non-truth-conditional criteria for DPMs. However, they do show that sådan has a referential function linked to its ‘standard’ meaning. As we explore later, this lends itself to textual and intersubjective functions when speakers use sådan in this way without referring to something present in the discourse. The DDO definition also signals possible subjective functions of sådan. Definitions (2b–d) above resemble a mitigation or hedge of the statement that follows it. These signal the speaker’s stance relative to the statement and/or provide the hearer with information on how to interpret the statement (i.e. not literally).

3.2 Previous studies on Danish DPMs and sådan

The seminal works of Torben Andersen (Andersen 1982) and John Andersen (Andersen 1986) first called attention to discourse-pragmatic functions of Danish adverbs, which Andersen (1986) termed sprogvidenskabens stedbørn [linguistics’ stepchildren]. Andersen’s (1982) point of departure is the modal particles da, nu, ellers, altså, and også. Modal particles for Andersen (1982) are adverbs typical of informal, spoken language that do not contribute to the sentence’s propositional content. Moreover, he argues, the modal particles help to structure discourse and facilitate positioning oneself relative to the interlocutor’s utterances and thoughts. Andersen (1986) similarly underscores what he calls refleksions-adverbialer [reflexive adverbs], which have functions that don’t relate to the utterance’s propositional content but rather reflect the speaker’s position towards the utterance. He outlines four types of reflexive adverbs, including distancemarkører [distance markers], which soften potentially categorical statements, and kriteriemarkører [criteria markers], which convey the degree of confidence in one’s statement. Although neither work investigates sådan, our paper shows clear overlap with the descriptions in Andersen (1982) and Andersen (1986) of (typically informal, spoken) particles that do not contribute to utterances’ propositional content, structure discourse, and allow speakers to express stance.

Since Andersen (1982) and Andersen (1986), there have been about two dozen studies on the Danish småord [little/small words]. For instance, studies have analyzed altså (Andersen 1982, Heinemann & Steensig 2018), da (Andersen 1982, Davidsen-Nielsen 1996, Krylova 2016), dog (Davidsen-Nielsen 1996), egentlig (Jensen 2006, 2009; Heegård 2015), ellers (Andersen 1982), godt (Jensen 2009), faktisk (Jensen 2006, Heegård & Mortensen 2014), jo (Davidsen-Nielsen 1996, Christensen & Jensen 2014), nok (Davidsen-Nielsen 1996, Sneskov 2008), nu (Andersen 1982, Davidsen-Nielsen 1996, Christensen & Jensen 2014), nåja (Emmertsen & Heinemann 2010), sgu (Davidsen-Nielsen 1996), skam (Davidsen-Nielsen 1996), så (Krylova 2016), vel (Davidsen-Nielsen 1996, Sneskov 2008, Jensen 2009, Krylova 2016), and vist (Davidsen-Nielsen 1996, Sneskov 2008), among others. However, sådan is left unanalyzed in these studies.

The same holds true for the largest works on Danish grammar. In Hansen & Heltoft’s (2011) introduction to Grammatik over det Danske Sprog [Grammar of the Danish language], the authors describe seeking to include neglected areas of study in their work. These include småord [little/small words] that feature prominently in spoken discourse, since det er uomgængeligt for en funktionel grammatik at få dem med i beskrivelsen [it is unavoidable for a work on functional grammar to include them in the description] (Hansen & Heltoft 2011:12). However, sådan is absent from Hansen & Heltoft’s (2011) sections dedicated to DPMs (they use the term dialogiske partikler [dialogic particles], see Section 2 above). Similarly, sådan does not make the list of ‘most common’ discourse particles in Lundskær-Nielsen & Holmes’s (2010) Danish: A Comprehensive Grammar.

The form sådan has been studied by association as a component of Danish quotatives (Rathje 2011) and general extenders (Rathje 2006; Christensen & Jensen 2015, 2018; Larsen 2015). Quotatives are markers of quotation that can report direct speech, internal thoughts or feelings, and introduce ‘vocalized sound effects’ (Levey 2006:414) and body language like facial expressions. In Rathje’s (2011) quantitative sociolinguistic analysis of the Danish quotative system she finds evidence of a quotative usage of være + sådan (glossed as ‘to be’ + ‘like’), as in (5). Indeed, 19% of all quotations involved one or more quotative particles, including sådan.

(5) det var bare sådan a::h

‘it was just like a::h’

(from Rathje 2011:72)

General extenders (e.g. ‘and stuff’, ‘or something’, ‘and things like that’) are phrases that typically generalize what comes before them to some larger set (Tagliamonte & Denis 2010). Researchers have found the constructions to have discourse-pragmatic functions, for instance in marking shared knowledge and performing politeness strategies (Cheshire 2007). In Danish as in other languages, general extenders (variously called løse udgange [loose endings], påhængsudtryk [attachment expressions], and generelle udvidere [general expanders]; Christensen & Jensen 2015:206) largely follow a prescribed formula. Larsen (2015) says a Danish general extender must have at least two constituents: a conjunction og/eller ‘and/or’ and a noun phrase. The noun phrase can consist of a quantifier (e.g. al ‘all’), a pre-placed comparative (e.g. sådan ‘like this/that’), a generic expression (e.g. noget ‘something’), and a comparative in the form of an adjective (e.g. lignende ‘similar’) or prepositional phrase (e.g. i den stil ‘in that way’). Christensen & Jensen (2018) report a similar formula, as shown in Table 2.

Table 2. Formula for general extenders in Danish from Christensen & Jensen (2018:82).

Note: The English translations of the Danish examples are ours and should be seen as ‘loose’. Especially when broken up into individual pieces, the words can be exceptionally difficult to translate.

Sådan is clearly involved in Danish general extenders, as in (6) and (7). Indeed, Larsen (2015) found og sådan noget ‘and something like that’ to be the most common general extender used in her data set, and Christensen & Jensen (2018) found sådan used in at least 73% of the general extenders they found.³

(6) jeg ved det ikke tror jeg sov # fyrre timer eller sådan noget lignende

‘I don’t know I think I slept forty hours or sådan something similar’

(from Christensen & Jensen 2018:208, our translation)

(7) de lyder som nogle der kommer fra Brøndby eller Ishøj eller sådan noget

‘they sound like people who come from Brøndby or Ishøj or sådan something’

(from Larsen 2015:81, our translation)

While these quotative and general extender studies touch upon sådan, they are not studies of the DPM form sådan as such. In the current paper, we treat the quotative and general extender uses of sådan separately from the DPM use for three primary reasons. First, while general extenders are commonly accepted as a sub-category of DPMs (Tagliamonte & Denis 2010), the literature is divided over whether quotative particles qualify as DPMs and/or whether quotative functions should be included alongside other functions in DPM studies (for studies that combine analysis of quotative particles and DPM uses, see e.g. Meehan 1991, Jucker & Smith 1998, Fleischman & Yaguello 2004. For studies that separate the two, see e.g. Ferrara & Bell 1995, D’Arcy 2005, Diskin 2017, Diskin & Levey 2019, Corrigan & Diskin 2020). Second, in both the quotative and general extender, sådan is required as part of a larger overall construction, not a standalone form. Third, our decision is consistent with the idea that individual quotatives and general extenders be examined in the context of their respective systems as a whole. (On quotative, see Tagliamonte & Hudson 1999, Buchstaller 2014, Davydova & Buchstaller 2015, Diskin 2017; on general extender, see Cheshire 2007, Tagliamonte & Denis 2010.)

To our knowledge, only three studies focus specifically on the Danish form sådan: Jensen (2013a, b) and Pedersen (2014). Jensen (2013a) focuses on the difficulties of assigning sådan – which she describes elsewhere as ‘highly polysemous and polyfunctional’ (Jensen 2013b:89) – into a word class. Drawing inspiration from Himmelmann’s (1996) classification, Jensen (2013a) categorizes three principal usages of sådan: (i) a situational deictic use with a visible/audible referent, (ii) an anaphoric or cataphoric use, where the referent is present in the preceding or proceeding discourse, and (iii) a recognitional use, where the intended referent is presumed to exist within the common ground of the interlocutors. She collapses her first and second categories into anaphoric (‘as shown/mentioned’) and argues that this anaphoric versus recognitional use (‘as you know’) is the fundamental differentiator in sådan’s function. The excerpt in (8) shows an anaphoric use, whereas excerpt (9) illustrates the recognitional use.

(8) Hvad udspringer en sådan mentalitet af.

‘Where does sådan a mentality stem from.’

(from Jensen 2013a:59, our translation)

(9) de seneste år har der boet mange forskellige mennesker i huset, men det var aldrig nogle, man sådan snakkede med.

‘the past few years there have been lots of different people living in the house, but they were never people you sådan talked with.’

(from Jensen 2013a:70, our translation)

Jensen (2013b) builds on this distinction by examining if prosodic patterns distinguish between the two functions (where anaphoric is stressed and recognitional is unstressed). Jensen’s (2013b) results lead her to conclude there is no correlation between sådan’s stress and its function.

In Jensen’s categorization of the recognitional sådan, we see clear links to DPM’s intersubjective functions, as detailed in Section 2. Himmelmann (1996:240) describes the recognitional use as ‘serv[ing] to signal the hearer that the speaker is referring to specific, but presumably shared, knowledge’. This knowledge can also adopt a sense of modality, such that sådan’s function is paraphrased as ‘du ved hvordan det føles/du ved hvordan det er’ ‘you know how it feels’ or ‘you know how it is’ (Jensen 2013a:70). That the speaker isn’t always certain if the referent is indeed mutual knowledge helps explain why confirming shared assumptions with questions (e.g. ‘You know?’, ‘Right?’ or ‘Remember?’) often accompany the recognitional use (Himmelmann 1996, Jensen 2013a). Jensen (2013a) concludes by acknowledging that she hasn’t analyzed all uses of sådan, but says the recognitional role seems dominant in certain contexts. Furthermore, these intersubjective contexts could clarify additional functions for sådan.

Samtlige mulige anvendelser og brugskontekster for ordet sådan har ikke været nævnt her – der er simpelthen for mange. Det skal dog nævnes at den anamnestiske kontekstbetydning kan bidrage til at anskueliggøre endnu flere anvendelser. Det ser for eksempel ud til at den kan blive den dominerende i visse kontekster. (Jensen 2013a:70)

‘All possible uses and usage contexts for the word sådan have not been named here – there are simply too many. It should however be noted that the recognitional contextual use can serve to illustrate even more uses. It appears for example that it [the recognitional use] can be the dominant one in certain contexts.’

Pedersen (2014) uses a conversation analytic approach to analyze 46 tokens of sådan in the construction ‘sådan + indefinite article (en/et/nogen/noget) + noun’ (henceforth ‘SÅN + en + N’ to align with her usage). Like Jensen (2013a), Pedersen distinguishes between the standard sådan that refers to something in preceding discourse and the sådan typical of spoken Danish, which is (i) often highly phonologically reduced, (ii) can never be declined as an adjective, and (iii) appeals to shared knowledge outside the discourse. Contrary to Himmelmann (1996), Pedersen (2014) finds the shared knowledge is not always incredibly specific. By showing examples where the referent is later called by a different name within the same general category of ‘things’, Pedersen shows the ‘SÅN + en + N’ can represent a prototypical example. She highlights that sådan is frequently used with ‘first mentions’ (Himmelmann 1996:236). First mentions occur when a noun/referent is first introduced into the discourse and are points where it can be unclear whether the noun/referent is indeed within the shared knowledge of both speakers. Like Jensen (2013a), Pedersen often observes sådan with appeal particles like ikk’ ‘right’, du ved ‘you know’, nok ‘I suppose’, and jo ‘you know/you see’ (translations from Davidsen-Nielsen 1996, Axelsen 2007), underscoring the active intersubjective nature of ‘SÅN + en + N’. In fact, Pedersen finds the preferred listener response is epistemic confirmation that the referent is indeed within the listener’s shared knowledge. This can be accomplished with an acknowledgement token like ja ‘yeah’, mmm ‘mmm’, or nå ‘oh’. Pedersen’s data shows no active disagreement on the shared knowledge, but rather says the listeners stayed silent to (presumably) signal their epistemic uncertainty. If the listener pauses instead of providing acknowledgement, the speaker would often continue elaborating or describing the referent in question.

In our paper, we expand on both Jensen’s (2013a, b) and Pedersen’s (2014) findings. Using a discourse analytic framework, our paper seeks to delineate the additional uses Jensen alludes to with the aim of giving a holistic picture of sådan’s multifunctionality. Drawing on Jensen’s anaphoric and recognitional concepts, we will study the nuances of sådan’s intersubjective functions and explore sådan’s additional textual and subjective discourse-pragmatic functions. As regards Pedersen (2014), we find that sådan functions intersubjectively outside the boundaries of the ‘SÅN + en + N’ construction. We further develop her observation of speakers continuing to elaborate after using the recognitional sådan. And finally, while her data set only includes silence as a signal of inaccessible shared knowledge, our corpus includes examples where listeners (inter)actively negotiate the shared knowledge in real time.

4. Method

4.1 Qualitative approach

As a preliminary investigation into sådan, this research joins the majority of DPM studies that are conducted qualitatively (Pichler 2010:602; Tagliamonte 2012:247) and focus on DPMs as features of discourse representative of ‘language-in-use’ (Cameron 2001:13). Qualitative paradigms grant researchers ‘important insights into [DPMs’] functionality, their context sensitivity, their syntactic, semantic, and prosodic integratedness as well as their evolution’ (Pichler 2010:602–603). Beyond such benefits, a qualitative analysis is in line with a bottom–up, inductive approach typical of discourse analytic approaches (Wood & Kroger 2000:34) which are concerned with ‘what and how language communicates when it is used purposefully in particular instances and contexts’ (Cameron 2001:13). A bottom–up approach is also more suitable when the feature under investigation is understudied (as is the case with sådan), and there are little to no guiding principles from previous literature, unlike, for example, work on English ‘like’, which has been extensively studied and can be coded based on functional taxonomies from previous research (see e.g. Diskin 2017). A qualitative approach is also appropriate for the present study given the modest size of the sample.

4.2 Form over function

DPM studies can use either form or function as the primary starting point of investigation. Pichler (2010) argues that defining variables by function, i.e. delineating features that serve to exemplify, draw attention, express approximation, etc. neglects the fact that (i) DPMs are ‘polysemic elements’ (p. 589) whose function may not be stable enough to define, and (ii) DPMs acquire new functions over time. Alternatively, a form-based approach sidesteps these issues by using form, being more stable and transparent, to investigate function. By concentrating on the form sådan, this research joins the majority of qualitative studies that look first to form to elucidate function(s) (Waters 2016:49).⁴

4.3 Pilot observations

SamtaleBank (MacWhinney & Wagner 2010) – a corpus of Danish talk and transcriptions – served to confirm our initial observations about sådan’s use. For the pilot study, we chose a two-party videotaped conversation entitled ‘Anne og Beate’.⁵ The discussion involves two women, likely in their twenties or thirties, recorded around 2010 (Johannes Wagner, personal communication, 30 April 2018).

There are 61 tokens of sådan in the 10-minute conversation. When analyzing the examples, we found several with DPM characteristics, as in examples (10)–(12).

(10) så glemte jeg at du havde haft ringet fordi at jeg var ved at være sådan rimelig vissen

‘Then I forgot that you had called because I was starting to get sådan pretty tipsy’

(lines 203–204)

(11) og så: øh for den ene han render sådan rundt å hele tiden skifter skjorte

‘And then ah the one he is scurrying sådan around constantly changing shirts’

(lines 269–270)

(12) altså normal kutyme tænker jeg bare hvis man kommer (.) som den eneste person ind i et selskab der overhovedet ikke kender nogen (.) så er det sådan social kutyme at man ø:h (0.4) ø:h sådan indlemmer vedkommende i det sociale [fællesskab ikke]

‘I mean I just think the normal custom is if someone comes (.) as the only person in a group who doesn’t know anyone (.) then it is sådan social custom that you uh (0.4) uh sådan include that person into the [group right]’

(lines 279–282)

In (10) sådan seems to underscore or emphasize that the speaker was ‘pretty tipsy.’ In (11) the speaker is in the midst of her narrative – defined as ‘oral versions of experience in which events are recounted in temporal order’ (Schiffrin 1981:959) – and sådan draws attention to the action. In the third excerpt, in (12), the two sådans could be interpreted as mitigating or hedging. If the speaker pronounced her view of the custom as the social custom, it could come across as expert-like. Sådan hedges the utterance to signal the statements as opinion rather than fact. These observations showed enough DPM potential to warrant collecting spoken discourse data.

4.4 Data collection

To collect the data, the first author approached pre-existing, organically formed groups in a university cafeteria and recorded the naturally occurring talk using the iPhone Voice Memos app. The first author sought to approach groups that: were speaking Danish, comprised between two and four speakers to keep transcription manageable, were seated to facilitate recording, and appeared to be similar in age to maintain an age-homogenous sample. Groups could participate for as long as they liked, resulting in conversation lengths ranging from five minutes to 56 minutes. All participants gave informed consent for their recordings to be used for research.

Thirteen of the 16 participants were female. Females had a higher willingness to participate than the male groups that were approached, and at the cafeteria it was estimated that 80% of diners were female. Of the transcribed conversations provided in the present paper, all ten participants were female. The ten participants ranged in age from 24 to 28 years old, with a median age of 26 (SD = 1.14). The breakdown of corpus participants can be found in Appendix 2 in the Supplementary Materials online.

The transcription protocol is roughly based on Du Bois et al. (1993). A noticeable departure is that, due to time constraints, we did not transcribe based on intonation units. This sacrifices a part of the prosodic information structure, while still allowing for the analytic operations required by this paper. Also, while we largely maintain standard Danish orthography, there is some ‘modified’ standard orthography (Clift 2016:51). Specifically, the phonetically reduced form of sådan is written as så’n following the DDO (see Section 3.1 above) and standalone instances of [ʌ] which could mean at ‘to’ or og ‘and’ are transcribed as å as in SamtaleBank (see example (11) above).

We transcribed only the first 15 minutes for conversations lasting longer than 15 minutes, given the laborious transcription process and sådan’s high frequency in the data. The four transcriptions were, respectively, five, eight, 15 and 15 minutes long (see Table 3). Following transcription, it was found that recording length correlated closely with number of words uttered in each conversation. The average words per minute across all conversations was 213. For full transcriptions, please refer to the Supplementary Materials online.

Table 3. Length of conversations which made up the corpus.

4.5 Data analysis

Coding was done manually by the first author and transferred into NVivo. Any subsequent updates to the coding were made directly in NVivo. Codes were subsumed into four categories: quotative, general extender, non-DPM, and DPM (for the literature precedent and our explanation for the three reasons we treat general extender and quotative differently from DPM we direct the reader to Section 3.2). Quotative was generally clearest, followed by general extender and non-DPM, though all involved what Hasund (2003:3) refers to as ‘fuzzy cases’. The categories were operationalized as described below. Those instances meeting Pichler’s (2013) definition and not included in the below categories were categorized as DPM. Section 5.1 examines the DPM evidence in detail.

4.5.1 Quotative

Quotative sådan helps to report direct speech, internal thoughts or feelings and introduces ‘vocalized sound effects’ (Levey 2006:414) and body language, such as facial expressions. As there was no video recording, body language couldn’t be analyzed directly, and we tried to ascertain it indirectly using contextual clues (e.g. periods of no speech with sound effects). Distinguishing between speech that reports internal thoughts (quotative) and speech that makes a statement (not a quotative) was difficult.

4.5.2 General extender

We relied on the formula for general extenders from Larsen (2015) and Christensen & Jensen (2018) as outlined in Section 3.2. A fuzzy area with general extenders is they can reduce to merely comprise og sådan or eller sådan (Larsen 2015) which can also overlap structurally with DPM usage, especially when the speaker is commencing an utterance.

(13) artists der sad og snakkede og sådan de var virkelig meget det der med, (…) de har alle sammen en notepad ved siden af dem

‘artists who sat and talked and sådan they were really into that kind of, (…) they all have a notepad next to them’

(C3:67-69)

In (13), one could argue that sådan elaborates the speaker’s description, drawing attention to what follows it. However, we interpreted it as an extensional category of what the artists were doing (speaking and talking and so on) and coded it as a general extender. In most instances of og sådan or eller sådan it was clear if it extended a previous idea or not.

4.5.3 Non-DPM

The least straightforward coding distinction was non-DPM versus DPM. As described in Section 3, sådan has an inherent comparative semantic meaning that has a textually connective or referential role baked-in to the traditional, non-DPM usage. We counted sådan tokens as non-DPM only if they referred to some preceding topic in the discourse. Sådan used in fixed phrases, such as sådan set ‘just about’ or sådan så ‘so that’ were also coded as non-DPM.

4.6 Limitations

The methodology has some limitations. First is the risk that we came with a priori assumptions about sådan’s behavior, as this paper stems from a Master’s thesis which suggested a similarity with English ‘like’. This connects to criticism that, since interpretations in qualitative DPM studies are subjective, the ‘analyst can find whatever it is that [s/he] is seeking’ (Beeching 2016:6). However, comparative DPM studies by nature set out to compare two or more DPMs (e.g. Park 1998, Aijmer & Simon-Vandenbergen 2003, Hasund 2003, Overstreet 2005). Future studies on sådan are the best ways to understand the extent to which a priori assumptions may have influenced our analysis.

Second, there is a chance that coding could have been inconsistent. Upon reflection, the first author sometimes saw a different possible discourse-pragmatic function for specific tokens, and no intra-coder reliability was performed, though it should be noted doing so is not commonplace in qualitive research (O’Connor & Joffe 2020). For one, we were not trying to quantify the levels of different functions within the DPM category. The four quantified categories were clearly operationalized (see Section 4.5 above). For another, since DPMs are multifunctional, the mere presence of several functions should not cause concern but rather show all functions potentially at work are being considered (Cheshire 2007, Waters 2016).

5. Results

5.1 Research Question 1: Is sådan a DPM?

In Section 5.1 we systematically review sådan against the features of Pichler’s (2013) definition. We show that DPM sådan is (i) syntactically optional, (ii) has little or no contribution to truth-conditional meaning, (iii) expresses stance, (iv) guides utterance interpretation, and (v) coheres discourse.

5.1.1 Syntactic optionality

Removing a DPM from an utterance leaves the utterance both grammatical and intelligible (Fraser 1988:22). Compare removing ‘like’ from these two sentences:

(14) It looks like a snail.

(from D’Arcy 2005:3)

(15) We were [like] ready to [like] mutiny.

(from D’Arcy 2005:6)

Doing so in (14) renders the sentence ungrammatical, whereas in (15) it remains grammatical following the removal of both instances. Here we test sådan against this criterion.

Below are three examples of sådan where removing it would make the statement ungrammatical. Removing sådan would turn (16) into a question (Do you use the thesis-writing area?) and make (17) and (18) ungrammatical.

However, there are many tokens of sådan where the word could be removed and leave the sentence grammatical and intelligible. This is true for each of the prolific instances of sådan in (1) at the beginning of this paper, but also in (19) and (20) below. The utterances remain syntactically intact without sådan.

In some instances, a structural feature of Danish grammar makes it easier to identify when sådan is functioning in its non-DPM, syntactically required form. In Danish, when something (e.g. time or place element, object, subordinate clause) is placed before the subject of the main clause, the Subject–Verb order flips to Verb–Subject. In Danish this inverted word order rule is often called inversion (Drengsted-Nielsen 2014:27). In (21)–(24), adapted from Lundskær-Nielsen & Holmes (2010:596), adding the time element ‘tomorrow’, object ‘the piano’, or adverb ‘maybe’ requires a switch to the Verb–Subject order:

The current data include examples of speakers not inverting the Subject–Verb order when the traditional non-DPM form would require it. In (25), Mette describes her frustration with a former workplace. She came up with ideas and initiatives that her employers put into practice, but she never received credit for them.

Note here the word order så’n – jeg – tror (Subject–Verb) and not så’n – tror – jeg (Verb–Subject). If Mette had used the inverted Verb–Subject word order, removing sådan would have made the sentence ungrammatical. The meaning would have been along the lines of ‘It is not even, I think I worked like that for three years there’, where the ‘like that’ refers to some previously mentioned way of working. Instead, this is DPM sådan. It elaborates on her story by exemplifying why she disliked her former workplace. Because sådan is a DPM it does not trigger inversion, so removing sådan in this case would leave the sentence grammatical and semantically intact.

In (26), Noora describes the yoga class offered at her student dorm. For her, a yoga teacher, it is at the beginner level, but she’s fine with that.

If Noora had inverted the word order (og – så’n – er – det), sådan would be its traditional, non-DPM form. Removing it would have turned the utterance into a question. The whole sentence would be incoherent given the lack of inversion in the second Subject–Verb pairing. It would be akin to ‘*And is it fine with me and it is sådan really chill and relaxed?’. As in (25) above, removing sådan in (26) would leave the sentence entirely grammatical and semantically intact.

In sum, DPM sådan displays syntactic optionality, whereas non-DPM sådan is syntactically required. The Danish grammatical rule of inversion provides an even more compelling delineation between the two usages than a language like English can reveal. These findings suggest Pichler’s (2013) syntactic optionality criterion is met.

5.1.2 Contribution to truth-conditional meaning

A DPM does not alter the conditions under which a sentence is true. Here we test sådan against this criterion.

In (27), Eva and Lis had just been discussing the idea of having a notepad next to one’s bed for when thesis inspiration strikes in the middle of the night. Eva then adds that she saw an Adidas talk where creative types recommended such a strategy.

Sådan here precedes the provision of an example of a new concept, but it is not referring to an established entity in the discourse. Rather it is a ‘first mention’ in the ‘SÅN + en + N’ formulation Pedersen (2014) describes (see Section 3.2 above). It is an appeal to (potential) shared knowledge outside the discourse, so sådan does not impact the propositional content of the utterance. These findings suggest the truth-conditional criterion too appears to be satisfied.

5.1.3 Stance expression

A DPM can act subjectively to express speaker stance, where the speaker positions him/herself relative to the interlocutor and topic at hand. Here we test sådan against this criterion.

In (28), Noora performs extensive discursive legwork before finally getting to the brunt of her utterance – her former co-worker was ‘bitchy’.

In prefacing this description, Noora goes out of her way not to come across as the type of person who freely maligns others. She describes the co-worker as ‘sweet enough,’ though only at times and not ‘especially often.’ When she finally verbalizes it, she uses sådan, together with ‘a little,’ to soften the statement further, followed by an appeal for understanding to her listener with ‘right’ (literally, ‘no?’). Sådan allows Noora to communicate her opinion of someone while simultaneously mitigating potential judgment from her listener. These results suggest Pichler’s (2013) subjective criterion is satisfied.

5.1.4 Utterance interpretation guidance

A DPM can act intersubjectively to guide utterance interpretation, where the speaker provides communicative clues to narrow the meaning potential of what is said and to manage the relationship between speaker and hearer. Here we test sådan against this criterion.

In the first of three selected excerpts, Mette answers Noora’s question about how her new workplace, an offshoot of a much larger and established retail brand, compares to the parent store.

Mette uses sådan to paint a picture for her interlocutor. Noora has never visited the store. The first sådan resembles a hedge – coupled with lidt ‘a little’ and agtig ‘ish’ – whereby Mette tries to evoke Noora’s notion of a concept store to help her grasp what the store is like. In other words, she appeals to a paradigmatic example – that of the ‘concept store’ – and simultaneously signals that this entity is roughly under the same category. With the second sådan, she elaborates, using sådan as an exemplifier to build upon her previous statement, a discursive move that is akin to adding brushstrokes to the painting. Like Jensen’s (2013a) recognitional use, Mette’s use of sådan guides Noora’s interpretation of the topic under discussion.

In the next example, the speaker uses sådan to appeal to and gauge the listener’s knowledge base. Here, Noora has just explicitly checked in with Mette’s knowledge, asking if Mette has ever been to Bali. When Mette says she hasn’t, Noora appeals to Mette’s potential knowledge of other locations Noora believes to have a similar vibe.

In (30), sådan functions similarly to (29) where the hedged, paradigm-appealing function is followed by elaborating, exemplifying usages. Noora uses sådan to describe and check understanding. Note also the cooperative act of Mette employing sådan to add to Noora’s description, even though Mette has never visited Bali. This excerpt illustrates the construction and negotiation of shared assumptions and understanding in interaction, thus highlighting some instrumental intersubjective facets of sådan.

In the final extract, sådan is used during metalinguistic commentary that explicitly guides utterance interpretation. Mette describes how she felt when she’d go to work at an unenjoyable former job.

Mette says her stomach hurt when she went to work. But, realizing Noora may interpret this literally, she includes a metalinguistic comment where sådan underscores she didn’t literally have a stomachache, but rather she felt a sense of dread. Thus, using sådan allows the meaning potentials of her statement to be explicitly narrowed.

In sum, the three excerpts illustrate how sådan functions intersubjectively to manage the relationship between speaker and hearer. The first two excerpts show intricate negotiations of Jensen’s (2013a) recognitional use (‘as you know’) and Brinton’s (1996:38) ‘confirming shared assumptions’ as the speakers appeal to and gauge the listeners’ knowledge base. The third excerpt, in (31), highlights sådan’s function in ensuring understanding, in this case whether something is to be understood literally or figuratively. These findings suggest Pichler’s (2013) intersubjective functional criterion is met.

5.1.5 Discourse cohesion

A DPM can act textually to furnish cohesion between utterances, resulting in discourse. Here we test sådan against this criterion.

Elaborating the speaker’s train of thought is perhaps sådan’s clearest use in creating discursive cohesion. In (32), Mette describes her student job. She enjoys the current job much more than a previous job which, at the excerpt’s start, she has just finished describing.

Mette appreciates a student job where she can mentally clock-in and clock-out. The three sådans here elaborate her statements about the value she finds in her job. Sådan, especially the last two, is a reformulation, akin to ‘in other words’ or ‘for example’ or ‘what I mean is’. In this way, sådan provides discursive connectedness and has an illustrative, explanatory function. These results suggest Pichler’s (2013) textual criterion is satisfied.

5.1.6 Summing up

In sum, we are on solid footing in both understanding and analyzing Danish sådan as a DPM as per Pichler’s (2013) definition. Evidence from the corpus shows sådan can function: 1) subjectively in expressing stance, 2) intersubjectively in guiding utterance interpretation, and 3) textually in constructing cohesion in discourse. DPM sådan is syntactically optional, and in some cases the rules of Danish syntax highlight DPM sådan’s syntactic optionality in ways not possible in English. Finally, as with DPMs more generally DPM sådan does not show an ability to alter truth-conditions.

5.2 Research Question 2: Functions of sådan in Interaction

In Section 5.1, we argued sådan does qualify as a DPM in line with Pichler’s (2013) operationalized definition. As part of our investigation, we showed how DPM sådan serves textual, subjective, and intersubjective macro-functions. In this section, we seek to specify sådan’s functions more granularly. The results show sådan illustrates; exemplifies; marks hesitation; approximates a quantity; mitigates, hedges, or softens; and allows self-correction or self-repair.

5.2.1 Illustrating (or expanding/explaining)

Sådan can be used to ‘illustrate’ (Diskin 2017) by expanding or drawing attention to certain elements in the discourse. In (33), what sådan highlights is an important factor in understanding the difficulty Clara has with recording her thesis interviews.

That she must do several things simultaneously – record a conversation, create a conference call through a special app, and conduct the interview – is what makes things difficult. The speaker underscores this fact by way of sådan.

5.2.2 Exemplifying

The data show speakers frequently use sådan to give an example, usually where glossing it with for eksempel ‘for example’ is possible. In (34), Noora has just mentioned she’s contemplated moving to Bali after graduation, with Mette suggesting it is a great time to do so.

Noora says ‘the job’ but then uses sådan to exemplify how the job is ‘freelance’ (and thus flexible) by listing examples of what ‘the job’ entails: freelancing by writing and making a documentary.

In (35) the speaker has just received a compliment for her minimalistic stroller setup. She says that she and her partner aren’t the types to bring lots of luggage with them. They even take only carry-ons for trips to the U.S. She concludes:

The bolded sådan gives an example of something that would not really ‘be us’ – namely, one of those big diaper bags that other parents carry around.

5.2.3 Filler or hesitation marker

Sådan is a useful linguistic resource when speakers hesitate or want to buy time to plan what to say next. In (36), Lis discusses her plans to visit the family summer house.

Sådan gives Lis time to formulate what she wants to say, in tandem with ‘so,’ ‘but,’ ‘just,’ and ‘I think’. It appears to be a ‘conventionalized or routinized’ (Hasund 2003:234) clustering of words without much independent import. It could be argued that this is a cluster consisting entirely of DPMs, as at least their English equivalents have been treated in the literature (e.g. for ‘so’ and ‘but’, see Schiffrin 1987; for ‘I think’, see Kärkkäinen 2003:178; for ‘just’, see Tagliamonte 2005). Eva seems to interpret the sådan-cluster as a filler or hesitation marker as well. She takes the opportunity to interject with a clarificatory question of her own.

5.2.4 Approximative adverb

In the data only one clear example of sådan as an approximative adverb was found. In (37), the speakers discuss prices for recording apps.

Sådan approximates the price 40 kroner, something reinforced given the speaker gives a price range for the apps.

5.2.5 Mitigator or hedge

Sådan as a mitigator or hedge was common in the corpus. Section 5.1.3 showed how sådan can be used to mold stance expression for politeness purposes. In the corpus, sådan was found to soften statements or assessments, especially when combined with lidt ‘a little/a bit’ (see examples in (28), (29), and several in (30) above). For example, in (38) Lis tells Eva how she feels stressed while writing her Master’s thesis.

With the first sådan, the combination with lidt ‘a little/a bit’, which occurs frequently when sådan functions in a mitigating/hedging fashion, reinforces a softening of her statement that she has been a sucky friend. It is a way of protecting herself from being overly critical or making a sweeping statement. The second sådan appears with another potential hedge, bare ‘just’. This second sådan could be interpreted as a hesitation marker, or as another hedge to avoid overstating how irritable she has been. (We discuss sådan’s overlapping functions in Section 5.3 below.) In both instances here sådan could be glossed with ‘sort of’ or ‘kind of’.

5.2.6 Self-correction (or replacement/repair)

While less common than illustrating or mitigator functions, the data also shows sådan used to self-correct or repair a previous statement. The ‘self-correction’ function can frame the correction of an error, misinformation, or a false start (Diskin 2017). In (39), the speaker describes the complications of recording while simultaneously conducting an interview.

Here sådan precedes a ‘self-initiated self-repair’ (Clift 2016:236). The speaker begins with the subject jeg ‘I’ (i.e. first-person singular) but decides to change it to interviewet ‘the interview’ (i.e. third-person singular) midway through the utterance.

In (40) one could argue Clara uses sådan as a filler. But given the lack of pause we believe a better interpretation is that she uses sådan to replace what she had started saying (det hele ‘the whole’) with a more specific formulation of what she meant (den der veksling ‘that shift’). The excerpt in (40), then, is another example of self-initiated self-repair with sådan.

5.2.7 Summing up

The extracts analyzed in this section reveal a breadth of sådan’s interactive functions. Based on the current data, sådan can serve: (i) illustrating, (ii) exemplifying, (iii) filling or hesitating, (iv) approximating, (v) hedging or mitigating, and (vi) self-correcting interactional functions. Some functions appear to have already been reported in the DDO definition (see Section 3.1). For instance, the mitigator/hedge function mirrors definition (iib), which says that sådan is ‘used to express that a statement approximates a more apt or precise formulation’, and definition (iic) which states that sådan is ‘used as an expression for hesitation or caution’. Similarly, the approximative adverb function aligns with definition (iid), which says that sådan means ‘Approximately, about’ and in the example sentence is used to qualify an amount. Sådan’s illustrating function appears to have some overlap with Jensen’s (2013a) recognitional use. Other functions, including sådan’s exemplifying, hesitation marking, and self-corrective functions do not appear to have been reported elsewhere.

5.3 Multifunctionality

In Section 5.2 we suggested a range of discourse-pragmatic functions that sådan can play and provided exemplars of those functions in interaction. However, our data reveals many instances of sådan involving multiple, simultaneous discourse-pragmatic functions. Because our paper is an exploratory examination of a previously under-described and understudied expression, we want to illustrate the wide breadth and rich interweaving of sådan’s functions. In the Discussion we use examples from this section to support the idea that assigning DPMs a principal function risks overlooking their multifaceted functional nature. In the interest of space, we have only narrowly included previously given extracts, and direct the reader to the appropriate sections of this paper for the fuller extract and additional explanation.

First, we turn to an ‘illustrating example’. In (41) Clara describes the mixed feelings she has about having to give a presentation in German.

It is notable that sådan accompanies the highly-planned (see Schourup 1999), contrastive, discourse-structuring pair ‘on the one hand… on the other hand (OT1H/OTOH)’. The discourse-marking pair can be seen as both playing a discourse-structuring (textual) role that establishes ‘coherence relations’ (Scholman, Rohde & Demberg 2017:47), and a marking, or illustrating, role (textual/intersubjective). Here sådan could be interpreted as a textually driven exemplifier, loosely paraphrasable with ‘for instance’. But it could also be drawing attention to the contrast that ‘OT1H/OTOH’ communicates. Fleischman & Yaguello (2004) argue against the idea that multiple consecutive DPMs serve redundant roles, asserting instead that each DPM is ‘presumably carrying out a different pragmatic function’ (p. 139). Even if this is true, it is difficult to tease out which of the two DPMs (sådan or OT1H/OTOH) is illustrating, exemplifying or discourse-structuring. This example demonstrates the rich interweaving of sådan’s interactional functions, especially when combined with other DPMs, and the difficulty of delimiting just one principal function.

Second, we show an ‘exemplifier-elaborator’. In (42) the second sådan demonstrates overlap between the illustrating and exemplifying functions, which we coded as ‘exemplifier-elaborator’. It could be interpreted as textual, offering a continued example that structures the discourse, or as an illustrator, elaborating on new details of a continued train of thought. Neatly delineating the two is difficult.

Third, we analyze a ‘hedged exemplifier which includes a tentative offer’. In (42), the first sådan demonstrates the overlap of a hedged exemplifier. The speaker adds to the ongoing description of the store, but here also mitigates the statement together with words like ‘a little’ and ‘ish’.

What we describe as a ‘tentative offer’ occurs in (43) and (44). The example in (43) gives a potential description of Bali, and (44) expands on the spirituality the interlocutor experienced at a yoga class. Both are statements offered by the listener to add, rephrase, or check her understanding of what the speaker is saying. As such, they demonstrate sådan’s interactive, intersubjective utility.

Neither (43) nor (44) have a question intonation. Question intonation would signal a difference in cognitive assumptions and make a request for clarification, akin to how ‘like’ can mean ‘you mean’ (as in: A: I’m with Christian. B: Like from my linguistics class?). Instead, they are said as statements, so are best understood as hedged exemplifiers. In neither case was the person present to experience what is described, and thus may be unsure if the interjected assessment matches what the interlocutor describes.

Fourth, we turn to a ‘paradigmatic example in focus’. In instances like (42), (45), and (46), sådan seems to contain hedging, exemplifying, and illustrating aspects. In all three, the speaker draws attention to the example which signals a reservation that the example is not precisely identical. Nevertheless, it is sufficiently representative to paint a picture in the mind of the interlocutor.

In (42), it is the notion of a concept store. In (45), it is the vibes of certain other beach cities. In (46), it is the concept of a Nørrebro apartment. The hedging aspect could signal the imprecise fit the example represents, but it could also acknowledge that the two speakers’ have different conceptualizations of a Nørrebro apartment or concept store and feelings about such cities as Miami or Venice Beach. Even if the speaker were to see a precise overlap between, say, Bali and Venice Beach, it is uncertain that the listener would agree. These three examples illustrate how sådan can evoke an example while simultaneously marking it as new, important information and mitigating the degree of its suitableness as a paradigm.

To sum up, the excerpts analyzed in this section underscore the densely interconnected functions that can co-occur within a given instance of sådan. They illustrate simultaneous: illustrating and exemplifying (also as elaborating and exemplifying), hedging and exemplifying, and combined illustrating, hedging, and exemplifying to evoke a paradigmatic example. The intertwined discourse-pragmatic functions exist both within and between the textual, subjective, and intersubjective domains.

6. Discussion

This paper has shown sådan’s noteworthy multifunctionality. Used in quotatives and general extender constructions, and to serve a diverse array of discourse-pragmatic functions, sådan performs a wide breadth of roles. DPMs have been described as ‘notoriously multifunctional’ (D’Arcy 2017:1), meaning they can not only serve different discourse-pragmatic functions in different contexts, but they can also exhibit several functions simultaneously (Andersen 2001:64; Beeching 2016:6), between or within textual and interpersonal modes (Schiffrin 1987:60–61; Hasund 2003:57; Pichler 2010). At times in the present data, flavors of different pragmatic functions – a dash of hedge, a pinch of exemplification – permeate the usages. Neatly delineating a specific instance’s principal discourse-pragmatic function from the multiplicity of sådan’s functions is difficult, and, as Cheshire (2007:188) would argue, counterproductive. As Waters (2016:51) adeptly notes, assigning instances to ‘single functional categories is in danger of giving one function priority over any other(s) that might also be at work’. Such tactics risk our understanding of the larger functional picture. Indeed, we have argued that sådan can simultaneously serve several functions. For instance, we have demonstrated overlap between the textual and interpersonal domains, such as when sådan precedes the provision of an example, playing a cohesive role, while simultaneously helping to mark the following information as important or hedged/mitigated, playing an interpersonal role (see Section 5.3). Qualitative, context-based discourse analyses of DPMs are valuable in the nuanced understanding they allow.

In addition to improving our general understanding of DPMs, the results presented here have implications for future research. First is the value of corpus-based dictionaries in more accurately reflecting actual language use. The corpus-based DDO seemed to have identified several DPM uses in their definition of sådan. Dictionaries can often be conservative in their definitions and editors can be loath to add what they deem ‘ungrammatical’, ‘dialect’, ‘slang’, or ‘vulgar’ usages (Andersen 2001:215–216), regardless of how widespread they are in everyday life. If dictionaries do include a definition that mirrors DPM usage, it is often maligned. For example, the prestigious Danish–English dictionary from Gyldendals Røde Ordbøger describes sådan as a tomt fyldeord [empty filler] before translating to the English ‘sort of’ (Axelsen 2007). And the Oxford English Dictionary describes its sole definition of DPM ‘like’ as ‘a meaningless filler’ (https://en.oxforddictionaries.com/definition/like; accessed 20 April 2018). In contrast, relying on corpora enables the DDO to follow its goal of focusing on moderne dansk sprog [modern Danish language] and more accurately reflect actual language use, including the use of sådan. Having more corpus-based dictionaries could provide useful benchmarks and reference points for future linguistic studies. Second is the value of studying a breadth of DPMs across languages to uncover generalities as well as idiosyncrasies. The Danish grammatical rule of inversion sheds some light on sådan’s DPM status (see Section 5.1.1). This example shows that grammatical rules of a given language can facilitate DPM categorization.

As regards future directions, firstly, this study did not include an analysis of sådan’s positional distribution. Previous DPM research has benefitted by showing the markers’ positional restraints, or relative lack thereof. Position in the sentence has been shown to correlate with discourse-pragmatic function (Diskin 2017). In Danish, Andersen (1982) suggests modal particles retain their ‘non-modal’ meaning if positioned at the beginning of the sentence (fundamentfeltet, see Diderichsen 1946), whereas their modal functions emerge when positioned within the sentence (nexusfeltet, see Diderichsen 1946). Both Davidsen-Nielsen (1996) and Lundskær-Nielsen & Holmes (2010:413–414) make similar claims, saying that Danish DPMs are restricted to the central field of the sentence. There is also evidence from Norwegian and Swedish that DPMs do not appear in initial position (Altenberg 2010) but can appear in either central or final position (Borthen 2014, Fretheim 2014, Berthelin & Borthen 2019). Future analysis of Danish DPMs’ position within the sentence could inform our understanding of their developmental trajectory (D’Arcy 2005:209), including their grammaticalization paths (Andersen 2001:272, 294).

Second, this paper has established that sådan is highly recurrent among the twenty-something female participants who comprise the transcribed corpus. A future sociolinguistic study could investigate usage across apparent time to put these participants’ use of sådan into context. Although the situation is complex, there is a popular view that young females are often linguistic innovators (Labov 2001:279; Tagliamonte 2005:1896; D’Arcy 2007:412, 2017:119; Meyerhoff 2011:225; Beeching 2016:11). So, if the frequency and function of sådan’s use among this study’s participants is broader than the general Danish population, such a study could establish whether there is a change-in-progress in sådan’s development into a DPM.

Third, studies could identify what other DPMs, if any, exist in Danish that play a similar role to sådan. DPMs rarely have just one equivalent across languages (Mosegaard Hansen 1998:7–8). One potential candidate is the Danish word ligesom ‘like, as’. If we look at (47), with both sådan and ligesom, it appears that ligesom also displays DPM characteristics. The speaker here is discussing her sister’s boyfriend, who is a musician and is also eager to have children.

Sådan works together with lidt ‘a little’ to hedge against condemning the boyfriend for having an entirely unrealistic expectation. Ligesom appears either to draw attention to the unlikeliness of the situation working out or soften the blow of such an assertion.

Ligesom appears in Andersen’s (1986) list of ‘distance markers’ (distancemarkører), which he suggests are used to soften what could otherwise be perceived as a categorical claim. And the Norwegian liksom (the corollary of Danish ligesom) has been studied by Hasund (2003), where she found it played very similar discourse-pragmatic functions to what we have found for sådan. Interestingly, Hasund (2003:243) reported that liksom tended to cluster with sånn ‘like this/like that’, which is the Norwegian equivalent of Danish sådan.

Frequency of ligesom in the current corpus (14 total tokens) is dwarfed by sådan (273 total tokens, 157 as DPMs). Moreover, the functional versatility of sådan appears to be larger than ligesom, at least for the moment. Nevertheless, it seems well worth a future research endeavor. For instance, if there are indeed two simultaneous ‘like’-related DPMs in Danish, is there evidence for a change-in-progress, whereby sådan as a DPM fully replaces ligesom? Or, during such competition between two functionally similar linguistic variants, will one of the forms specialize into a narrower functional niche? Both options are for instance in line with Fruehwald & Wallenberg’s (2013) general theory of categorical linguistic variation and could prove an interesting test case if the variation between sådan and ligesom is investigated and tracked over time.

Fourth, if it is determined that sådan shares functions with ligesom/liksom, future studies could determine if sådan constitutes a Danish parallel to the English DPM ‘like’ which shares a lexical source with ligesom/liksom. Research has attested to ‘like’-related DPMs in languages as varied as Bislama, Finnish, Italian, and Hebrew (Levey 2006). Fleischman & Yaguello (2004) argue that ‘like’-related DPMs across languages are found in lexical forms that originally possess a semantic meaning of comparison or similarity, which is clearly true in the case of sådan. Based on their findings of extensive overlap between English ‘like’ and French genre, Fleischman & Yaguello (2004:143) issue a call-to-action: ‘Cross-language pragmatics would be well served by additional studies testing out the findings of this investigation on data from other languages’. Such studies first identify the markers used by different languages – a goal to which the current study contributes – and then sketch the functional-semantic and syntactic development stages (Fleischman & Yaguello 2004). Such cross-language study drives at a core question of DPMs: ‘To what extent do all languages share a basic set of [them] with the same core pragmatic meaning?’ (Fraser 1990:395).

7. Conclusion

While DPMs have been described as showing ‘semantic shallowness’ (Brinton 1996:34), they are not meaningless and in fact ‘pertain to the very essence of communication’ (Wierzbicka 1986:519). Indeed, after having confirmed Danish sådan as a DPM, this paper has illustrated the myriad important discourse-pragmatic functions sådan displays in interaction. In service of larger textual, subjective, and intersubjective macro-functions, sådan is used in the ten female speakers’ 43 minutes of transcribed conversation to illustrate; exemplify; mark hesitation; approximate a quantity; mitigate, hedge, or soften; and self-correct or self-repair. Moreover, we have shown that sådan operates multifunctionally, often incorporating several functional ‘flavors’ at once. These results were possible because of the form-based, qualitative approach taken to understand the full range of functions on display in each instance of sådan. Such an approach also problematized the privileging of one functional category at the exclusion of others. This study’s value extends to its inclusion in the relatively smaller DPM literature on languages other than English. It has demonstrated the importance of studying DPMs cross-linguistically, as with the Danish rule of inversion that demonstrated sådan’s grammatic optionality. Maschler & Schiffrin (2015:205) write that even studies focusing on a relatively narrow aspect of DPMs’ meaning can ‘teach us something about their role in discourse’. The goal of this paper has been to serve as a preliminary, qualitative, form-based investigation into the functions of DPM sådan in hopes of promoting further qualitative and complementary quantitative studies on sådan specifically and Danish DPMs generally.