1. Introduction1

The use of okay in spoken discourse has been described in various languages and interactional settings. Its functions cluster around two main features. First, okay can mark a transition between two chunks of talk or activities; second, okay is said to agree with a previous turn or information. While the former use appears to be frequent in institutional, monologically oriented settings, the latter frequently occurs in less asymmetric encounters. Regarding the second use, there is no consensus in the literature on the exact action that okay carries out, as it is said to display for instance understanding, agreement, or compliance. In this contribution, I will try to tackle this conceptual vagueness by exploring how occurrences of okay in German are linked to interactional issues of understanding, repair, and revision of prior assumptions. Based on a small data set of video-recorded everyday encounters in German, I will explore its most frequent responsive uses in a selection of unconstrained face-to-face conversations, i.e., as a stand-alone token and combined with a change-of-state token such as ah or achso. With the help of a qualitative, micro-analytic approach (conversation analysis) I would like to argue that okay itself is not related to claiming understanding but conveys a more neutral epistemic position. By using okay, the speaker claims acceptance of an information that they are not able to verify or to confirm, either because of a lack of direct epistemic access to prior information or because this piece of information runs counter to their displayed and/or formulated assumptions. More specifically, by okay participants can claim non-agreement, i.e., a neutral position with respect to prior information.

In what follows, I will first present the main studies of okay in oral discourse in different languages (2.1), and, more specifically, in German (2.2). I will then briefly present the data and method used for this contribution (2.3). After a description of the distribution and forms of okay in my data set (3.1), the analytic section will elaborate on the link between okay and change-of-state tokens in general (3.2) and, more specifically, between okay and ah okay (3.3). The analysis will finally present some cases of stand-alone okay and reflect on differences between falling and rising intonation patterns (3.4): while through okay the speaker always claims an acceptance of prior information without claiming a specific position towards it, okay with rising intonation more strongly points out an absence of agreement or a need to obtain more information than okay with falling intonation. In conclusion, okay is analysed and understood as a neutral and local response token that participants use in order to claim acceptance of – but not agreement with – a specific and local piece of information in the preceding turn (4.).

2. Okay in spoken discourse

This section will focus on research on okay in spoken discourse in various languages and in German; more specifically, it will comment on research within the domains of conversation analysis and interactional linguistics. On the one hand this is due to the analytic approach chosen in this contribution, on the other hand other contributions to this special issue contain more details both on the diachronic dimension of okay and on additional references in other types of approaches (Col & Delahaie, this issue).

2.1. Okay in various languages

The following overview will be structured according to the two main functions or uses of okay according to existing interactional and discursive research, first, the structuring of discourse, second, agreement. The interactional settings in which okay has been studied vary considerably (e.g., everyday encounters, medical settings, business meetings, service encounters, for an overview see Gaines, 2011, p. 3291-3292). On the linguistic level, mainly American and British English data have been considered (or other Englishes, cf. Adegbidja & Bello, 2001; Lee, 2017), although ongoing research aims at remedying this lack of linguistic variation (Betz, Deppermann, Mondada & Sorjonen, forthcoming).

First of all, okay has been identified as an important structural marker in oral discourse. Especially in institutional, monological talk such as university lectures (e.g., Looney, Jia & Kimura, 2017; Othman, 2010; Schleef, 2005, 2008), but also in other forms of teaching and counselling (e.g., Guthrie, 1997; Lee, 2017; Rendle-Short, 1999), okay – along with other lexical items – has been identified as a resource for structuring the talk, both for the speakers and their recipients. Okay can mark transitions, especially between larger chunks of talk or between different activities (so-called “vertical” transitions, cf. Bangerter & Clark, 2003). In phone calls, okay is frequently used as a device to work towards a closing (Schegloff & Sacks, 1973), or, in the beginning of phone calls, to move to the reason for the call (cf. Beach 1993, and, within another paradigm, Bangerter, Clark & Katz, 2004). In institutional talk such as medical interviews, okays can help doctors to steer topical development and to make their coparticipants adhere to the institutional task at hand (Beach, 1995a; see also Antaki, Houtkoop-Steenstra & Rapley, 2000).

Beach however underlines that okays are more than simple discourse markers, as these tokens are “positionally active and consequential for unfolding talk” (Beach 1995b, p. 122). He describes okay in American English as a typically free-standing response token that functions as an interactional pivot, because it responds to prior talk, but also projects a transition to next-positioned matters (Beach, 1993, 1995b). Although this transitional use has been described in earlier studies (e.g., in service encounters, Merritt 1980, or in decision taking, Condon, 1986), Beach complains about an analytic underspecification of the uses of okay (Beach, 1995b, p. 129), as previous studies don’t give detailed accounts of how okay works in a “dual fashion” (Beach 1995b, p. 139) in that it helps to close some prior action and thus enables the participants to project and move to a next matter.

Second, while most studies focus on the structuring and transitional value of okay, its functioning with respect to agreement or understanding is less explicitly investigated. Condon & Čech (2007) note that the accomplishment of a transition simultaneously marks the achievement of common ground, therefore okay displays understanding. More frequently however, agreement is treated as being the core meaning of okay (cf. Gaines, 2011; Schleef, 2005). Okay is often put in a same group with other response tokens that supposedly mark agreement (such as yeah, right, exactly, for English cf. Othman, 2010, for Hebrew cf. Maschler, 2002, for Greek cf. Pavlidou, 1998) or, more specifically, compliance (together with English alright and sure, cf. Thompson, Fox & Couper-Kuhlen, 2015, p. 225-233). More rarely, okay is analysed with respect to precise sequential positions or environments. Stevanovic & Peräkylä (2012) describe the Finnish okei in second, responsive position as a compliance token. Okei accepts the prior action, more specifically, “with ‘okay’ and ‘all right,’ the second speakers acknowledge that the first speakers’ assertions have consequences for them” (Stevanovic & Peräkylä, 2012, p. 305). According to Schegloff (2007, p. 120), okay “may mark or claim acceptance of a second pair part and the stance which it has adopted and embodied within the sequence.”. Contrary to oh, okay does not play a central role in informing sequences (Schegloff, 2007, p. 123), this is also shown by Seuren (2018) in his study on action ascription in third position in Dutch. Okay-prefaced deontic assessments figure frequently in arrangement-making sequences: while these assessments accept the previous turn as a proposal (for setting up some joint project or meeting), the okay more specifically acknowledges the prior action.

Overall, while the transitional value of okay seems to be acknowledged by various studies and frameworks, other functions of okay are described in rather eclectic ways, ranging from acceptance, agreement, and confirmation to compliance or understanding. A forthcoming cross-linguistic study of okay (Betz, Helmer & Deppermann, 2018; Betz et al., forthcoming) aims at providing a more clear-cut distinction by restricting the analysis to a specific sequential type of okay and by considering its generic sequential uses; on the one hand receipting and understanding, on the other hand transitioning and closing.

2.2. Okay in (spoken) German

According to standard German dictionaries (such as the Digitales Wörterbuch der deutschen Sprache (henceforth DWDS) or the Duden2), okay can be written both as a full lexical form (okay) or in its abbreviated form (O.K., OK or o.k., ok). In the Duden, okay has three possible lexical entries which are usually attributed to colloquial language: first, it can be used as a noun (with the neutral article, “das Okay”) with the meaning of ‘consent’, ‘approval’ or ‘agreement’, and in German, one can “okay something” (sein Okay geben, literally ‘to give one’s OK’). Second, it can be used as an adjective, used with a sense of ‘being all right’ or ‘good’ (Ger. ‘gut’), referring to someone’s character, someone’s physical state or, more generally, a state of affairs. In its adjectival use, okay also exists in aviation jargon and can confirm a piece of information after it has been checked. A third, adverbial form of okay refers mainly to possible discursive uses. It is supposedly used to express that something has been arranged, settled or agreed upon, in sentences such as okay gehen wir ‘okay let’s go’ (cf. Duden online3). The DWDS distinguishes only two entries of okay: okay used as a substantive or as an interjection, both being mainly linked to agreement and consent. This standardized view on the lexical meaning resonates with one of its discursive functions as it has been described by linguistic research, namely agreement. Standard dictionaries however do not report on possible closing or structuring functions as they occur in spoken interaction.

Linguistic research, on its part, investigates okay in German both as a structural marker and as a response particle. According to Werlen (1984), okay figures among the group of “summarizing particles” (resümierende Partikeln) together with also (‘so’), ja (‘yes’), and gut (‘good’). This closing function is visible for example in its use in German phone call pre-closings (Pavlidou, 1998). Okay has also been described as segmenting bigger turn units produced by one speaker, such as in university lectures (Schleef, 2005, comparing American English and German) or business meetings (Barske, 2009), in which it is used for structuring the activity and, in case of work meetings, facilitating its progression. In German maptask interactions, okay is frequently used in order to close a step within the overall activity (Proske, 2017, see also Bangerter & Clark, 2003 for tangram data in Swiss German). Okay has also been described in turn final position, in which it works as a tag question, along with other tokens such as stimmt’s (literally ‘is it right/correct’) or klar (lit. ‘clear’) (cf. König, 2017, p. 235). A comprehensive study of okay in German is provided by Helmer, Betz & Deppermann (forthcoming): it illustrates that a more fine-grained analysis of okay has to take into account possible variations with respect to the position of okay within a sequence, a turn or a turn-constructional unit; with respect to possible different intonation patterns; with respect to possible pre- or post-positioned particles; and, finally, with respect to the participants’ embodied conduct in the vicinity of okay. The authors show for instance that – in connection with different multimodal gestalts – okay with falling intonation can be used for displaying (sufficient) understanding or for closing larger activities. In German (and in other languages, e.g., English or Polish, cf. Betz et al., 2018) okay can also carry rising intonation. This can indicate a discrepancy between the speaker’s expectations and a piece of information contained in the prior turn (Betz et al., 2018): okay? treats this information as unexpected or insufficient, and it is not sequence-closing.

Overall, the study of natural interactional data shows that okay in German does not simply or only relate to agreement, as normative descriptions of this lexical item usually point out, but that – at least in naturally occurring interaction – it is also related to the structuring and closing of activities, sequences or topics, and, more interestingly, to the claiming of (sufficient) understanding of a previously delivered information, especially if it carries falling intonation. The analytic section (3.) aims at investigating how precisely okay in German relates to interactional matters of (non-)understanding of and of responding to prior information.

2.3. Data and method used

The data used in this contribution stems from video-recordings of everyday interactions that have been collected between 2014 and 2016 in different parts of Germany. All participants are native German speakers from different regions and speak standard German, sometimes containing minor dialectal, lexical or phonological elements specific to their respective native region. For this analysis of okay, three recordings have been selected:

• Corpus BYB, a family reunion at home with four participants (length: 110 minutes)

• Corpus PHEVE, an outdoor evening chat among four friends (length: 60 minutes)

• Corpus PHPIZ, an outdoor evening chat among 3-5 friends (length: 45 minutes)

The data have been transcribed according to the Jeffersonian conventions (Jefferson, 2004a). All proper names (except for makes of car, e.g., excerpt 7) have been replaced with pseudonyms. As a supplementary intonation mark the semicolon has been used for mid-falling intonation (see Selting Auer, Barth-Weingarten, Bergmann, Bergmann, Birkner, […], Uhmann, 2009). This leads to four different intonations: “.” for falling, “;” for mid-falling, “,” for mid rising and “?” for rising intonation. A smile voice has been indicated by “☺”, with pointed brackets that delimitate the concerned elements of the turn. Selective multimodal annotations have been made according to Mondada’s conventions4 (cf. Mondada, 2013, 2018). In the transcripts the original talk is in black font, the translation to English is in blue, the multimodal annotations in grey. The turn that initiates the sequence leading to the use(s) of okay is marked with a grey background, the responding turns containing okay are made visible by using a turquoise background.

The analysis is based on ethnomethodologically grounded conversation analysis (Sacks, Schegloff & Jefferson, 1974; Schegloff, 2007; Sidnell & Stivers, 2012), an approach that considers the precise timing and formatting of speaking turns in order to describe systematic patterns of naturally occurring social interaction. More specifically, the following analysis will consider not only audible, but also some visible aspects of the social encounter, in order to take into account how the participants use talk, laughter, gaze, gesture and other embodied resources for managing the interactional jobs at hand (a so-called multimodal approach, e.g., Goodwin, 1981; Mondada, 2007; Streeck, Goodwin & LeBaron, 2011; Deppermann & Streeck 2018).

3. Analysis

In this part I will try to tackle how the use of okay in German is linked to the interactive negotiation of knowledgeability, expectations, or understanding. More specifically, I will show that – at least in the available data set – okay is most frequently used for accepting prior information but claims neither agreement nor confirmation. Also, this acceptance does not necessarily relate to the understanding of prior information. By responding with okay, the participant claims having less or no knowledgeability about it, which is also shown through their previous questions regarding the matter at hand, or because it counters their previous (and sometimes explicitly formulated) assumptions.

The data overview will illustrate (3.1) that okay most frequently occurs as a free-standing token in the selected data set. If combined with other tokens, okay most frequently occurs together with change-of-state tokens such as ah and achso. Moreover, even if okay occurs as a free-standing token, change-of-state tokens frequently appear to be used by the same participants in the sequential vicinity of okay. The analysis therefore evolves around a contrastive analysis of different instances of okay and cases of okay combined with pre-positioned change-of-state tokens. The notion of change-of-state token has been initially coined by Heritage’s studies (1984, 1998, 2002) on English oh, by which a speaker can propose a change in their current state of knowledge; it is typically (but not exclusively) used to respond to completed informings. While German ach and achso are often treated as being equivalent to English oh (for an overview see Golato & Betz, 2008, p. 8-9), it has been shown that speakers use these items for different purposes. By achso, participants propose that they have received and understood new information, by ach however, they treat some piece of information as received, but not necessarily as understood (Golato & Betz, 2008). Although German has a large variety of change-of-state tokens (cf. Betz & Golato, 2008, p. 60), okay seems to particularly well combine with achso and ah. During the analysis, more research on German change-of-state tokens will be quoted when relevant.

In order to illustrate that speakers of German use different tokens for claiming that they understood prior (new) information vs. that they accept it, this section will reflect upon the proximity of okay to change-of-state tokens (3.2). More specifically, I will show how ah okay (in contrast to okay alone) is used to index a revision of prior assumptions (3.3). Finally, the analysis will seek to demonstrate that by using okay, the speaker claims being unable to verify or to explicitly agree with the previous information, and that okay possesses a limited sequential scope (i.e., okay works as a “neutral” and “local” acceptance token, 3.4).

3.1. Overview: quantitative description of okay in the selected data set

As okay in spoken German is rather frequent, the data set that has been chosen for this contribution has been deliberately limited. In the nearly 3,5 hours of conversational data, more than 100 occurrences of okay have been found (see table 1). Different types have been distinguished: first, okay only, i.e., without other response tokens (mainly used as a stand-alone token, i.e., as single element of a turn); second, okay in combination with a pre-positioned change-of-state token (Heritage, 1984) such as ah, achso, aha and ahja (all in one intonation unit); third, in combination with other response tokens (such as ja ‘yes’ or mhm_hm); fourth, cases in which okay is not used as a response token, i.e., either as part of reported speech or together with the copula be (“that’s okay”).

The numbers regarding the different occurrences of okay aim at giving a global impression of the overall frequency and distribution of okay in a sample of non-experimental, informal multi-party interaction in German. The numbers and percentages represent estimates with regard to the selected data set and do not replace an in-depth qualitative analysis of single occurrences. Nevertheless, the table is useful for sketching some general tendencies within the available data. In general, okay mostly represents an independent intonation unit, i.e., it is either stand-alone or clearly separated from other response tokens. Thus, in case of response token + okay one should rather consider that okay can co-occur with other, mostly pre-positioned response tokens (as there usually is a pause or clear prosodic break between the initial token and okay, indicated by (.) in table 1). 46, i.e., around 43% of the occurrences represent okay only, of which the majority have a falling intonation (only 11 carry a clearly distinctive mid-rising or rising intonation). A next big group (30 out of 106, or about 28%) contains the combination of a change-of-state token with okay. There usually is no pause between these tokens, the most frequent combination being ah okay and achso okay. A smaller group (15) is composed of combinations with various, but prosodically independent response tokens, note that these are overwhelmingly pre-positioned (e.g., ja ‘yes’, nee ‘no’, or echt ‘really’). The last group shows non-responsive uses, i.e., in which okay is part of reported speech (“and he was like (.) okay”) or part of an idiomatic and sedimented expression (is(t) okay ‘that’s okay’). Consequently, okay in this data set is predominantly used in response to a prior turn and has a specific affinity with change-of-state tokens. Turn-internal/final and non-responsive uses are clearly rarer, also, exclusively transitional uses of okay (structuring a single speakers multi-unit turn) could not be found. While in the selected data set, the prototypical use of okay is as a free-standing or at least as a turn-initial response token, a larger data set will be needed in order to see if this observation applies to ordinary, non-institutional talk in German in general.

3.2. Uses of okay related to issues of change-of-state

From the small data set used for this contribution, 30 cases of okay were preceded by a change-of-state token such as ah (15) or achso (11). Although these were a bit less frequent than the free-standing okays (i.e., without change-of-state tokens, 46), the analysis will start with the presentation of the former. As I will argue, the proximity of okay with interactional issues involving a change-of-state allows for a thorough understanding of its uses, precisely as it co-occurs with a change-of-state token that takes on another interactional task. The first excerpts will also illustrate that okay is often clustering in longer, sequentially and topically related stretches of talk (all of which is invisible in the bare numbers of table 1, and which will be relevant at a later point in the analysis, cf. section 3.4.).

In the first excerpt, René is commenting on Caroline’s son, who is playing with other children next to the table where the participants are seated, and who seems to look rather pale. In his question to Caroline, René offers two possible reasons for the child’s complexion, first, that his mother put suncream on him, second, that he simply is pale (01-02). René will receipt the following explanations by Caroline and Susanne by various forms of okay.

Both Caroline and Susanne (who co-explains throughout this sequence) choose the former option, i.e., that the boy’s whiteness is due to sunscreen (04-05). René responds to this by an achso okay, by the achso indicating that he understood the previous information (06, achso being frequently used in third position after confirmations of candidate understandings, cf. Golato & Betz, 2008, p. 32). When Caroline expands her turn and adds that she put cream on the boy’s face in the morning only (while the conversation has been recorded in the evening, i.e., more than eight hours later), this is responded to by ja ‘yes’, followed by a free-standing okay. Interestingly, while this bit of information could be new to René (the precise time and number of the creaming), it is not receipted with a change-of-state token. Instead, about 0.2 seconds later, René formulates a new question based on this information: as the sunscreen is still visible, he infers that it must have a high sun protection factor (SPF), fifty (08). This is however disconfirmed by Caroline and Susanne, as the whiteness is due to the type of sunscreen, a mineral one (i.e., a sunblock rather than a suncream), not a conventional chemical one that fully penetrates the skin and leaves no white trace once applied. Renés ah okay (13) in third position points to the receipt of new information, however, he does not seem to claim full understanding (see also the slightly rising, thus possibly interrogative intonation on okay, cf. section 3.4). Note that in this case ah okay acts as a third in a corrective sequence (see Caroline’s turn-initial “nee” ‘no’, 10 and 16). While Golato & Betz (2008, p. 32) actually describe ach + repeat as thirds of corrective sequences (and not ah + okay), their analysis could actually hint at the fact that okay does something specific in the way it receipts a prior turn, as in my data there are no repeats of prior information following the ah okay. Caroline and Susanne will continue their joint explanations regarding the suncream beyond Rene’s receipt (15-23). The sequence will then be abandoned due to some children coming to the adults’ table, so that René’s understanding (or absence of it) of this state of affairs will not be mentioned anymore.

This excerpt shows how a participant handles different pieces of information separately and differently. Achso okay seems to claim an actual understanding of the previous information or action, as there is no self-initiated expansion by Caroline or Susanne after its occurrence (06). Ah okay (13), as can be shown by the co-participants’ expansion of their answer afterwards (15-23), seems to relate to an absence of understanding. Finally, the okay in l. 08 illustrates that less important (or more detailed) information, i.e., that the creaming happened only once in the morning (07), is not receipted by a change-of-state token.

Okay is frequently involved in sequences where a speaker’s state of knowledge or their assumptions are verified or corrected. The second excerpt will provide an example of a free-standing okay that is preceded by a change-of-state token by the same speaker earlier in the sequence. Here, okay clearly relates to some information in a prior turn, but in itself does not seem to display the participant’s evaluation of the information as particularly new or as (not) understood. Nicki is telling a story about how she initiated a friend to motorcycling. As this friend was inexperienced, she explains how she was driving behind him and slowing down other road users in order to keep her friend safe. Elena first assesses this safety measure as good (04), but then latches an inference onto this assessment (06).

Elena infers that it was the friend’s first ride; the pre-positioned ach indicates the receipt of new information but makes no claims regarding the achievement of understanding (Golato, 2010). Indeed, Elena does not formulate the following turn (06-07) as an assertion, but with an interrogative format (with the verb in initial position). The incremental “quasi” (‘almost’) and the tag question (“oder was” ‘or what’) also qualify Elena’s assumption as not (yet) fully precise and to be confirmed by Nicki. Nicki does not directly disconfirm, but delivers a partial and modified repeat, thus providing an embedded correction (Jefferson, 1972) of Elena’s assumption, as it was not her friend’s first, but his second ride. Right after the modified element (“zweite” ‘second’, 08), Elena produces a stand-alone okay in overlap (09). The mid-falling intonation indicates possible sequence closing, and indeed Elena seems to treat the topic of the degree of inexperience as closed, as she does not comment on this anymore in a possible next position, and neither does Nicki (see Ludger’s turn, 11). Consequently, this okay seems to simply accept the modification of Elena’s prior assumption. As through the formatting of her question (06-07) Elena has claimed a reduced knowledgeability (Heritage, 2012) about the reported facts, okay cannot be used for confirming Nicki’s turn. Reversely, this means that okay could indeed enable a speaker to explicitly claim their non-confirmation or non-agreement (cf. the following analyses).

3.3. Ah okay vs. okay: indexing the revision of a prior assumption

While Golato & Betz (2008) and Golato (2010) analyse the pair ach vs. achso, there have been no instances of ach okay in the present data set. As ach “is typically used turn-initially and extends the current sequence” (Golato, 2010, p. 149), this projection of extension might possibly clash with the closing value of okay (cf. the global result that okay is often stand-alone, i.e., not immediately followed by more talk and thus not extending a sequence). In that sense, ah might be more neutral with respect to the projection of a possible extension than ach. According to Zifonun, Hoffmann, Strecker, […] Vorderwülbecke (1997, p. 386-387), ah expresses a speaker’s surprise or astonishment (see also Weinrich, 1993, p. 837-838), and indexes a positive contrast with respect to the speaker’s expectations. Imo (2007, p. 19-20), in his analysis of some cases of ah ja / so / o.k. underlines that due to this positive contrast, a sequence after ah x is usually closed, i.e., the speaker has no more questions and the information is treated as sufficient (cf. Betz et al., 2018, and section 2.2.). As the level of sequential granularity and the type of data used for the up to now quoted studies largely differ, these descriptions can only be treated as an analytical starting point and might even be slightly contradictory. The stand-alone ah in spoken German has not yet been studied, but one could consider some findings of vocalic change-of-state tokens in other languages. In French, by producing an ah-prefaced repeat of a responding turn, a speaker can index their own previous action as inadequate (Persson, 2015). In Finnish, “the token aa indicates that the speaker understands here and now something relevant that conflicts with his or her previous assumptions or that was otherwise unclear.” (Koivisto, 2015, p. 111). In Dutch, a free-standing oh claims that prior information was unexpected in that it contradicts prior assumptions or claims of knowledge (Seuren, 2019). This points to a possible core meaning of ah revolving around the revision of some prior element (concerning an action or, related to that, a specific assumption). German ah clearly belongs to the realm of understanding displays, more specifically, it could relate to a “now-understanding”, a revision of a prior assumption (cf. René’s ah okay in line 13, excerpt 1). While it won’t be possible to fully elucidate the precise differences between achso, ach, and ah in this contribution, the different actional features of ah (or other vocalic tokens) that have been mentioned here (treating information as unexpected, reversing assumptions, indexing now-understanding) will be helpful for understanding the complementary actions that ah + okay vs. the stand-alone okay can carry out.

In the next excerpt, Ludger is talking about his ideas for a trip to a North Sea island together with Nicki. As a fan of detective novels, he suggests visiting the spots on the island that have been mentioned in a specific detective novel, here, the place where “die Leiche” ‘the corpse’ (02) has been found. The definite article shows that Ludger alludes to a precise corpse (i.e., in a novel they talked about in the previous turns). This link however is not readily available, as can be seen by Nicki’s repair initiation regarding the facticity of this corpse (04): she adds a prepositional clause that states the “location” more precisely, i.e., the corpse has been found at this place “im krimi”, i.e., ‘in the detective novel’.

Ludger confirms Nicki’s candidate understanding with a literal repeat, framing the corpse more explicitly as belonging to the realm of fiction (see the laughter particles in 06 and at the end of 03, his smile voice and his comment “nicht in echt” ‘not for real’ afterwards, 08). In overlap, Nicki responds to Ludger’s confirmation with ah okay (07). Interestingly, in this case, Nicki explicitly formulates that she was assuming a different state of affairs, as she “thought that there was a past” (07, 09, thereby proposing an “in-principle correctness” of her assumption, cf. Jefferson, 2004b) to the location on the island, i.e., that it once was a real crime scene. The second, stand-alone okay in her turn seems to respond to the second part of Ludger’s turn “not for real”, 08). The fact that she then appends a next question (“und der autor?” ‘and the author?’, 10) shows that the repair sequence is now closed (although it is then expanded by Elena, who continues on the humorous tone, 11-12). The ah seems to relate to Nicki’s “wrong” expectation (as formulated by her immediately after the ah okay), as the second time (10) she does not reproduce the ah. As the repair has already been carried out (by the confirming “im krimi”, 06), the next bit of information delivered by Ludger (‘not for real’, 06, 08) is neither new nor contrary to her now-changed expectations. The stand-alone okay therefore seems to simply provide an acceptance of a second piece of information but does not specifically claim anything with respect to some prior expectation or with respect to a specific understanding. As the first syllable of the okay is lengthened and as Nicki withdraws her gaze from Ludwig at this moment (beginning of 10), she orients to a closing of the sequence about the corpse (cf. her looking back at Ludwig only on the article “der” – which introduces a new topic, the author of the novel, her gaze then wandering to Elena who addresses her in overlap).

The second excerpt of this section illustrates also how ah okay relates to bits of information that are new to the recipient or contrary to their expectations. More specifically, it is the ah that seems to respond to the newness or unexpectedness of the previous information: like in the previous excerpt, a second occurrence of okay produced by the same speaker is stand-alone and not accompanied by ah. Here, Elena asks Nicki about her new flat and how she gets along with her new flatmates. Nicki, after a first global and positive assessment (03), starts with an elaboration on the first flatmate, Claudia, who is a musician (05). The fact that Claudia plays the violin at home is receipted by Elena with a facial expression that displays a rather negative assessment.

While the first piece of information, the name and profession of the first new flatmate, is received by means of a neutral continuer (06), Elena receipts the next piece of information (that the flatmate practices playing the violin at home, 07-08) by a concurrent embodied assessment. She withdraws her gaze from Nicki and gazes to the left, raises her eyebrows and lowers the corners of her mouth (fig. 2, 08, compare fig. 1). Consequently, Nicki suspends her turn and responds to this facial display with a strong positive assessment (“mega schön” ‘mega beautiful’, 10, see also her responsive hand gesture fig. 3-4, possibly aiming at putting an end to Elena’s negative response, cf. Kamunen, 2018). In the meantime, Elena has turned her gaze back to Nicki, but has kept her negative facial expression (fig. 3). This display is dissolved shortly afterwards (fig. 4), when she responds in overlap with ah okay and some laughter particles (11). Nicki’s positive assessment (see also her repeat and explicit account of her flatmate’s musical skills, 12-13) illustrates her interpretation of Elena’s facial display as a negative assessment. Thus, Elena’s ah okay responds to a revision of her own expectation (e.g., that she expects the violin music to be a nuisance or of rather bad quality). Whereas the initial ah treats Nicki’s assessment as unexpected information, the okay seems to carry out a different interactional job. This can be seen in Elena’s following double okay (14, pronounced in one intonational unit, aiming at stopping Nicki’s positive assessment and account, cf. Stivers, 2004, see also her hand gesture in fig. 4, cf. Kamunen 2018). The okay itself seems to claim a neutral acceptance of some prior piece of information: first, as only Nicki seems to be knowledgeable about her flatmate (both Elena’s initial question and Nicki’s detailed elaboration on each flatmate indeed orient to Elena’s lack of direct access to this information), Elena is not in position to confirm that the music is indeed beautiful and that the flatmate has excellent musical skills. Second, Elena does not agree with Nicki either: this is also illustrated by the laughter particles (11) and her joking, alternative completion (17, Günthner, 2015; Lerner, 1996) of the compound noun “master of perfor(mance)” in Nicki’s turn (15-16). Elena’s version contains a rather negative undertone, compare her “master of disaster” with Nicki’s “master of performance” (16, which refers to an official title of a degree her flatmate aims to obtain). Also, Nicki’s elaboration on her flatmate’s professional music training (15-16) could in itself orient to the absence of agreement from Elena’s side.

If ah treats the newness or unexpectedness of a piece of information with respect to a prior expectation or assumption and if okay can’t be readily used for confirmation or agreement, okay must accomplish a different interactional task. In the next section, some examples of stand-alone okay illustrate its use as a neutral acceptance token, i.e., a token that claims (not more than) acceptance of some prior piece of information.

3.4. Okay as a neutral and local acceptance token

In this section I would like to argue that in spoken German, a participant uses okay in order to claim that they accept a piece of information while simultaneously claiming that, from their point of view, it cannot be proven, verified, confirmed or agreed with. This capacity of confirming or verifying a given piece of information can of course not be linked to a state of inner, cognitive, objectively “available” knowledge (cf. Heritage, 2012). However, I would like to argue that by responding with okay, participants publicly claim their impossibility to confirm or verify the veracity of the information, while at then same time displaying that they nevertheless give credence to what the previous speaker said. Okay allows the participants to not claim responsibility regarding a specific piece of information and provides for a neutral – or no more than – acceptance. In excerpt 5 Susanne has just talked about her sister-in-law who lives in the city of Jena. By his question, René displays that he most likely doesn’t know her as she hasn’t been present at any joint activities of their group of friends (02-03, 05). In what follows, Susanne mentions some possible joint acquaintances and possible events where René could have met her sister-in-law.

In the beginning of her response, Susanne turns to Annika, asserting that Annika’s husband Marc knows her sister-in-law (06-07). This additional information however does not lead to joint recognition, as Annika simply receipts this information by a free-standing okay (instead of, e.g., confirming it with genau, Oloff, 2017). By expanding on Susanne’s prior turn, René then suggests that Marc knows the sister-in-law because both originally are from the same city, Jena (13). Simultaneously, Susanne provides a next possible meeting opportunity between René and this woman, one of the past joint barbecue parties (12, 14-15). While Susanne asserts that her sister-in-law was at least once at a barbecue party, the meeting between René and her is presented as a possibility only (‘it could very well be that you came across her’), a fact about which Susanne is explicitly not certain (17-18). The only matter of fact that Susanne asserts is that her sister-in-law has been participating in at least one bigger party (18-19). None of the three participant seems to be more knowledgeable than the other. First of all, René – who should know best if he has met someone or not – explicitly prompts for more information in order to know if the sister-in-law is known to him (02). Second, by not confirming, Annika backs down from a possible K+ position attributed by Susanne’s turn (06-07, as she might know best about her husband’s acquaintances, cf. Raymond & Heritage, 2006). Susanne, for her part, cannot tell if René knows her sister-in-law, but can only assert that there was at least one event where René could have met her. René finally fails to recognize the person despite Susanne’s description, as illustrated by the absence of any change-of-state token (that could claim a “remembering” of the talked-about person or event, for example by using achja, Betz & Golato, 2008). His consecutive use of okay (16, 20) therefore does not confirm anything, but accepts twice something that he is not able to verify. The re-orientation of his head (ahead and thus away from Susanne, 19) shows that he projects a sequence closing and no further elaboration.

If we take a look at the intonation used on the okays in excerpt 5, it is interesting to note that René uses twice a falling intonation, while Annika uses a clearly rising intonation. Indeed, Susanne’s turn in 06-07 (to which Annika’s “okay?” in 09 responds) is presented as a confirmable (see the rising final intonation), thus, a type-confirming response would be a confirmation. Moreover, the fact that Susanne continues with a delay and while still being oriented toward Annika (11-12, also the other participants Isabelle and René have momentarily turned their gaze to Annika) displays that a more elaborate response from Annika has been projected. Through the rising intonation on okay, Annika more strongly distances herself from a possible knowledgeability attributed by Susanne’s turn, especially as she does not nod (e.g., in response to Susanne’s head nod, 07). René, for his part, provides two head nods, once after his first okay (17), then co-produced with the second okay (19-20). One can thus assume that an okay with falling intonation is more aligned and closing relevant than an okay carrying a rising intonation (but other prosodic cues could also play a role, cf. Selting, 1996).

In the next example, the participants are talking about their plans to do some karting together. Svenno introduces the name of a city (Hederow) as a possible place to go and adds that he forgot to send this information to his friends (01-02). Andi responds in different ways to this proposal (03, 09).

In overlap, Andi initiates repair regarding the city’s name (03). This confirmable is responded to by Svenno’s “genau” ‘exactly’ (05). Then Svenno explains that an outdoor kart circuit is located there and appends a positive assessment of this circuit, thereby displaying his knowledgeability (07, ‘really cool’). Andi responds to this with a free-standing okay (09). As Andi seems to know neither the city nor the circuit, neither a second assessment nor an agreement with Svenno’s assessment (cf. Pomerantz, 1984) is a possible next action for him. Thus, okay is accepting some prior piece of information or action without claiming any specific degree of knowledgeability. Here, neither Andi’s nor Svenno’s turn allude to any specific prior expectation from Andi’s side (contrary to the excerpts in the previous section, 3.3), which is why no change-of-state token is used at this moment.

Interestingly, in the next turn Svenno seems to generalize and therefore to downgrade his previous assessment, as in a referentially vaguer way he states that “und da auch- irgendwie:; (0.2) kannste auch (deinen spass haben.)” ‘and there [at the kart circuit] one can also somehow have fun’ (10-11). The vowel lengthening and the rising intonation on Andi’s okay (09) possibly reinforce Andi’s non-agreement with Svenno’s assessment and, at the same time, project an expansion on the presented information (see also the expansion of the participant having presented the information receipted by okay in excerpt 4). Thus, while the okay itself claims neither agreement nor disagreement, a rising intonation seems to push the okay more towards a possible disalignment – it receipts some piece of information as being in the domain of the other speaker and as impossible to verify, but, by doing so, at the same time more explicitly claims that it is impossible to agree with. It is not surprising that – if several okays occur within a sequence – rising okays have the tendency to occur in the beginning, while falling okays occur towards the end of the sequence.

The previous excerpt 4 (BYB_violine) provides another good example of this pattern: the first rising okay is responded to by Nicki with an argumentative upgrade (12-13, “wirklich” ‘really’, and the explicit mentioning of her flatmate’s skills that are “richtig gut” ‘really good’). The following okays – that carry a falling intonation – are not responded to in the same, upgraded fashion. Afterwards, Nicki elaborate on her flatmate’s qualifications, introduced by the conjunction “und” ‘and’. On the one hand, this shows that the okay with rising intonation is treated as more challenging, in that the first speaker seems to explicitly adjust the following action (e.g., add more information, reformulate a previous assessment etc.). On the other hand, the fact that Nicki continues elaborating on her flatmate’s musical skills beyond Elena’s second okays (15-16) shows that Nicki does not treat the subsequent falling okays as providing some kind of agreement neither. Instead of linking back to her prior, suspended – and indeed finally abandoned – turn (08, possibly relating to the relaxing effect that listening to violin music at home has on her), Nicki continues her response to Elena’s embodied assessment for a few more turns (explaining her flatmate’s different professional activities and career plans) before finally turning to the description of the second flatmate (not shown).

The expansion on the prior topic by providing additional information or adding new aspects following okay can also be observed in some of the other excerpts, see ex. 1 (13 ff), ex. 5 (09 ff), ex. 6 (09 ff). As the issue at stake is not systematically agreement or understanding, it is difficult to state what kind of response the first speaker exactly pursues by providing additional information beyond their recipient’s okay. However, one can state that:

• okays are frequently involved in long courses of action that elaborate on a precise matter of fact,

• okays with rising intonation receive more marked and explicit expansions or turn developments than those with falling intonation,

• while okays are used in responsive position and can thus close a sequence, the fact that the topic is not closed shows that they seem to sequentially act on a rather local level, i.e., they close a step within a longer course of action.

The last excerpt shows how okay provides a receipt of very local, precise bits of information as parts of a longer informing process (which can then later, and rather independently, lead to possible understanding, agreement or more explicit alignment). In excerpt 7, the participants are talking about different car models that existed in the former German Democratic Republic (GDR); here, they positively assess the Moskvitch, a Soviet-Russian car brand (01-04). While Svenno, Patrick and Markus agree and seem to have the same level of expertise with regard to this type of car, Andi presents his assessment in a less assertive way (05). His assessment that the Moskvitch was a ‘really cool limousine’ is presented as a confirmable (see the final tag ne, Jefferson, 1981). The others disconfirm this and thereby correct Andi’s assumption (10-15).

Andi’s assumption that the Moskvitch was a big, limousine type of car (05) is corrected by the others, as in their opinion this description applies to the GAZ Volga, which indeed was a limousine type of car. After having received a triple disconfirmation (10-15), Andi asks for more information about the alternative car, the Volga (16, if it was “noch- noch- (.) drü- (.) noch ne nummer;” ‘even a size bigger’), which is promptly delivered by his three co-participants (17-24): they claim that Volga was indeed bigger than the Moskvitch, that it was the official state coach and “der volvo (0.4) des:: (.) der DDR.” ‘the Volvo of the GDR’, i.e., among the most expensive and high-quality cars (22). Andi now responds to these multiple versions with a free-standing okay (25). Then, two more things happen: first – and as it has been described for other excerpts in this contribution – the co-participants continue elaborating on the description of the car (Patrick even uses his smartphone and projects a quotation from a webpage, 28); second, Andi formulates a follow-up question (29-30). This question (if Volga is an own make of car, i.e., indeed different from the Moskvitch) shows that his previous okay has not been used for claiming understanding, see also his frowning acting as a continuous display of repair initiation (fig. 1, 31) and his pursuit of confirmation (37). In what follows (not shown here), the three expert participants discuss different features of the two car brands, and at a later point Andi will formulate a candidate understanding that further enables him to identify both cars as belonging to different marques.

While an explicit account of understanding won’t occur in this long stretch of talk (and can’t be reached as the issue at stake for Andi is rather to learn about the precise differences between both marques), one can nevertheless state that Andi’s follow-up questions display that he did not yet have a full grasp of the different features and uses of both car models at the moment he utters the okay. The okay merely locally accepts the information previously presented by his co-participants (i.e., it responds to information in a directly preceding turn) but does not close the overall task of precisely identifying the features of the Moskvitch vs. other car brands.

4. Discussion and conclusion

While in the literature, okay in spoken discourse is related to various functions such as agreement, confirmation, and the display of understanding, the analytic section of this contribution aimed at obtaining a more precise and unified description of its role. The quantitative overview of the analysed data set has shown that okay is frequently used as a stand-alone response token or frequently occurs together with change-of-state tokens like ah and achso (see also Helmer et al., forthcoming). The detailed, qualitative analysis of some of its occurrences has shown that prototypically, okay itself:

• is not used for confirmation, as the participant using it explicitly displays an absence of knowledge regarding the information through prior requests for confirmation or repair initiation (indeed, okay frequently appears in third position of corrective sequences),

• is not used for agreement, as the participant is in no position to agree with something to which they don’t have epistemic access or which is contrary to their (often explicitly formulated) expectation,

• is not used for indexing a change-of-state, as this – when relevant - is indexed by pre-positioned change-of-state tokens, typically ah or achso.

Instead, okay is used to claim acceptance of some prior piece of information and enables the speaker to adopt a kind of neutral, “non-agreeing” position towards the given piece of information (see also Lindström’s, 2018 analysis of Swedish okej as an “agnostic” marker). This opens up the possibility of continuing the sequence in two different ways after okay; first, there can be an expansion on the same topic, i.e., the participant who delivered the initial piece of information adds more details or modulates their previous turn; second, the sequence can be closed and a new sequence and possibly topic is initiated. This means that okay in itself does not initiate repair. On the contrary, despite a possibly indexed reduced knowledgeability or revised prior assumption, okay does not project a (specific) next action. Although okay is generally stand-alone and therefore closing implicative, the fact that co-participants mostly expand on the matter at hand illustrates that the prototypical okay is indeed not treated as an agreement. These expansions could also be related to the very local scope of okay, i.e., it responds to the previous turn or sequence, but not necessarily to the larger scope of the ongoing talk. Okay often refers to some precise, local element within longer sequences and allows the participants to continue on the same topic, of which some mutual understanding can be reached at a later point. By using okay, a speaker does explicitly not adopt an evaluative position towards the previous turn but claims a “non-agreement”. This neutrality of okay for receipting information and its local scope could also explain why it lends itself so well to doing closings of bigger chunks of talk (for a similar reflection on the development of the responsive genau ‘exactly’ towards more monological uses see Oloff, 2017).

While the majority of okays in the selected data set carry (mid-)falling intonation, the cases of okay with (mid-)rising intonation appear to be particularly interesting. Indeed, a possibly interrogative intonation seems to reduce the neutrality of okay. In the cases of okay, or okay? presented here, the rising intonation hints at a possible disalignment, or rather, it hints at a possible disagreement or possible extended repair without actually projecting it, i.e., it is only minimally disaligning. By adopting a different intonation or by adding some vowel lengthening, German speakers seem to be able to bend the neutrality of okay to different degrees of “repairability”. In order to investigate the precise link between different prosodic formats of okay, more systematic and detailed analyses (e.g., by using the software Praat) will have to be carried out.

Another point to be investigated more thoroughly is the embodied conduct during and in the vicinity of okay. For reasons of space, the present contribution could not consider, e.g., the presence of head nods (or their absence, as both types of conduct can be observed in the selected data set). One could expect that the degree of “neutrality” of okay also depends on the type of concurrent embodied conduct. As different embodied actions (such as head nods or bodily reorientations) and their link to different uses of okay are currently investigated (Betz et al., forthcoming; Helmer et al., forthcoming), more cross-linguistic findings can be expected in the near future. Last but not least, the diversity of okay formats and the complexity of the sequences in which it is involved illustrate that this token is far more than a linguistic quirk: the frequency of okay in contemporary German shows its ability to fit a precise interactional task better than any of the already available response tokens.