Nonlexical “Moans”: Response Cries in Board Game Interactions

ABSTRACT This article examines nonlexical vocalizations in board game interactions, focusing on “moans.” Moans are prolonged, voiced, response cries. Moans react to game events where the player has suffered in some way. Despite the complaint-relevant nature of moans, game actions are never withdrawn in response to a moan, Moans are treated as laughable, while lexical complaints invoke arguments and apologies. This article suggests that moans are a manifestation of managing Bateson’s play paradox in that they denote suffering but also willingness to continue play and a validation of the prior event. Moans are suggested to be a contextualization cue for “this is play.” Given the relative unconventionality of the form of moans, these tokens are suggested as evidence that lack of conventionalization may be a members resource rather than a problem. The article analyzes a corpus of 34 hours of video-recorded board game play (169 tokens) in English (Canadian, American, and British).

as "the most obvious manifestation of nonseriousness" (Holt, 2013, p. 71); however, laughter is also bound up in other interactional tasks such as indexing delicate actions (Potter & Hepburn, 2010) and managing troubles resistance (Jefferson, 1984). Although laughter is common in the data corpus, the moans themselves and turns containing moans do not contain laugh particles, and yet participants do not treat moans as serious indicators of suffering. A moan (in the sense of the response cries described in this article) "does not carry the 'serious' consequence it would otherwise" (Glenn, 2003, p. 137); the relevant response is not empathy or rectification.
Other interactional studies of play have focused on children (e.g., Goodwin, 1985) and families (Fatigante, Liberati, & Pontecorvo, 2010). Despite the common use of board game sessions as data for interactional studies (e.g., Fox & Thompson, 2010;Kendrick & Drew, 2016), most studies of play, rather than other interactional phenomena, examine video game interaction (Mondada, 2012;Reeves, Greiffenhagen, & Laurier, 2016;Tekin & Reeves, 2017; but see two exceptions in Liberman, 2013;Hofstetter & Robles, 2019). Aarsand and Aronsson (2009) analyze response cries that occur in family video game play, finding that the broad category of response cries including lexical and nonlexical tokens are used to display engagement in the games (p. 1567). Goffman (1978, p. 791) also suggested that response cries demonstrate continuing attention and engagement, adding, "We owe, to any social situation in which we find ourselves, evidence that we are reasonably alive to what is already in it-and furthermore, to what may arise, whether on schedule or unexpectedly." Prior research into nonlexical vocalizations has also suggested a connection between response cries and establishing attention and/or displaying engagement. For example, Keevallik (2018) has shown how one function of strain grunts is to manage the timing of physical effort, such as timing the transfer of lifting a heavy weight between parties. Goodwin (1996) has shown how response cries can draw attention but also frame the triggering event as visible and understandable by co-present parties. Several studies have examined how nonlexical vocalizations coordinate attention to in situ experiences of pain (Heath, 1989;Jenkins & Hepburn, 2015;Weatherall, Keevallik, & Stubbe, 2019). The "flooding out" nature of these sounds, as predicted by Goffman, may contribute to their effectiveness in situations where the trigger for the response is very proximal; Wiggins (2002) found this to be true of tasting sounds like mm, saying they constructed immediacy and spontaneity.
Nonlexical vocalizations have also been connected with displays of affect. For instance, two affectual categories that are relevant for this article, given the occurrence of moans after a new game event, are surprise (Wilkinson & Kitzinger, 2006) and disappointment (Couper-Kuhlen, 2009). However, the tokens in the corpus do not match the way affect has been constructed in prior studies (first, they have different form-to-action attributes, and second, they are treated as nonserious), which provides further evidence that there is no one-to-one relationship between affect and prosody (Local, 1996). Reber's (2012, p. 245) account of affect in nonlexical vocalizations is particularly comprehensive, and she recommends viewing these sounds as evoking "prismatic meaning clusters which are methodically left unstated by participants." In other words, the less conventionalized nature of the sounds permits interactional work that lexical tokens do not.
In the ensuing article, I will describe the form of moans, I will analyze the sequential relevancies that lead up to and follow moan productions, and I will discuss how the moan sound makes nonseriousness relevant. This analysis will further our understanding of how play comes to be constructed, as well as contributing to the growing literature (e.g., Dingemanse, 2018) around nonlexical vocalizations.

Data
The analysis is based on a collection of 169 instances of moan sounds. The instances are collected from a corpus of board games between adult players, most of whom are hobbyists who regularly play board games, with a few new players who are the hobbyists' acquaintances. The corpus includes competitive games and collaborative games, with a variety of different game mechanics. There are 20 different games, over 26 sessions, with 55 different adults. Sixteen hours are from fieldwork at gatherings and board game cafes, 4 hours are from university students trying out games, and 14 hours are from games uploaded to the Internet. Most, but not all, were native speakers of English (true across data subsets). Participants in the first two subsets gave informed consent, and Internet recordings are treated as public television, given that they have between 10,000 and several million views each (note that online games were not scripted). The data collection was reviewed and approved by the ethics board at Loughborough University.
The claims about phonetic properties in this article are based on what Walker (2013) calls "auditory phonetic analysis" in that the sounds are analyzed with repeated listening and not with acoustic measurement. Acoustic analysis was not chosen as a large proportion of instances occurred in complete overlap, making it difficult to distinguish acoustically between speakers' productions. Unmotivated looking noted the phenomenon of responding with a nonlexical vocalization to events where the player suffered a loss, where their deception was unmasked, or where luck did not go their way, and a collection of 190 instances made by looking for nonlexical responses to events where at least one game player suffered in some way from a game event (i.e., some action was unsuccessful and/or the event made it less likely for them to win). A total of 169 of these tokens involved production of prolonged glides between various back or central vowels, between open-and midclose ([ɐ], [ɑ], [o̞ ]), often with epiglottal trills, described in this paper as "moans." The remaining 27 tokens were a mixture of unvoiced sounds (sucking air between teeth or sighing ahhhh) and nasals; these also receipted suffering game events, but often more "minor" ones, and few had the full moan sequence described in the following. A further eight victory cheers were found (from competitors), which involved front close vowels ([u] or [e]). The following analysis also draws on a collection of 90 complaint sequences involving lexical complaints and a variety of "lexical" response cries (swear words, and "No!"s). The complaints collected involved both instances of players responding immediately to game events and to events occurring at more distal, earlier moments in the session.
With respect to transcription, I follow the Jefferson and Mondada transcription systems. In the Jefferson transcript, I append the letter n to indicate nasalization, and gh to indicate epiglottal constrictions. The International Phonetic Alphabet symbols for the moan sounds appear in the line above the Jefferson transcribed sound, with volume indicated in subscript (e.g., f forte f ) bracketing the IPA. Unlike usual IPA, no square brackets are used, to avoid confusion with Jefferson overlap square brackets. Note that epiglottal fricatives may contain trilling.

Analysis
In the analysis, I will first describe the moan sounds in terms of form and sequential position and compare them to prior studies of similar nonlexical sounds. Next, I will elaborate how moans are treated sequentially by players. I will finally compare the moans to instances where game events are receipted with lexical tokens or lexical complaints.

Form and sequential slot of moan sounds
Moans were produced as multivocalic, using central and back, and open to midclose vowels (the loudest and most prolonged vowels in any given moan were produced close to one of [ɐ], [ɑ], or [o̞ ]). There are also other prosodic cues that regularly "cluster" (Selting, 2010) with moans, namely, prolonged duration, creaky voice, and low and/or falling pitch contour. Whether something is a moan is a member's problem, and different combinations of features produce similar treatments in the data corpus, with certain features clustering together. This article aims, in part, to demonstrate that the flexibility (or unconventionality) of moan production is one of its virtues, rather than some sign of trouble or incompletion.
Moans occur after a game event, issued by the suffering player. Moans appear to occur at what Selting (2017) calls "affect climaxes"; players have built up to the game event, and although the outcome was not projectable, the fact of an event occurring was projectable, and now players have a chance to react. This creates a slot where players can display their understanding of the event, its relevance for them and the game, and their personal reaction (such as affect or stance), and thus create congruent public versions of the game events (Goodwin, 2018). In the following instance, Tina has just completed a game turn. A new card is revealed (as one is at the start of every next game turn), and it is one Tina would have preferred to play, had it been available when she took her game turn moments ago.
Extract 1: GN Kodama_0:58:00 1 Tina moans on seeing the new card (line 3), which would have been a better option for her. Frank treats this as laughable, and Stephanie asks whether that card was what Tina would have wanted. Tina's utterance could have also been a realization of an alternate move or anticipation of Frank being able to make an especially good move. However, Tina ratifies that it would have been a better card, although not a hugely better card (she earned two points, rather than the three she might have gotten). This latter utterance also downgrades how serious the suffering was for her, contrasting to her initial moan that proclaimed the injury. Stephanie co-constructs Tina's moan to jointly produce a specific, congruent version of the flipping the card event and its consequences for Tina.
The next extract involves a speed game. When Adam says "blitz" (line 3), he ends the round, preventing players from placing any other cards or earning any further points.
Extract 2: 180815 Dutch Blitz_0:8:03 (see Figure 1) After Adam ends the round, and after a silence during which the other players examine the cards, Gwen and John vocalize. It often takes a beat of time for the players to visually survey the event and ratify what the consequences are. Gwen cuts off her vocalization, but John gives a moan sound. He states he was under the impression he was doing well in this round, implying that this impression is now broken, since he has many unplayed cards. Adam treats this as laughable (line 14). In Extract 3, the players have just passed a new set of cards to their left. They are now seeing what new possibilities they have for their next play. Ben and Hal appear to be addressing each other, whereas Tina's utterance is not directed at anyone through gaze and does not receive a response.
Extract 3: GN Sushi Go_0:18:35 As Tina flips to a new card, she produces a moan. She says a different move (on her previous placement) "would've been betterh" (line 2), now that she sees what her options are. After a micropause she downgrades this problem as dismissible, with "oh well." With this small set, we can report the following: First, players ascribe meaning and action to the sounds, such as in Extract 1 when Stephanie treats the moan as a display of regret. Second, the sounds can be produced with different vowels among the open vowel set. Third, players can overlap these sounds without interjacent overlap (a clearer example is in Extract 7). Fourth, the moans are initiated in a separate TCU and pitch contour from prior talk, though they may be latched to subsequent talk (see Extract 2). Fifth, most have a low and/or falling pitch contour (see Figure 1). Sixth, the sounds are prolonged. They are more prolonged than vowels in other syllables in the game data but also more prolonged than the nonlexical response tokens in telephone corpora like CallFriend, Newport Beach, or CallHome (see examples in Local, 1996, which examines the prosodic and phonetic characteristics of oh response tokens). Yaeger-Dror (2002, p. 1635) report that 200 ms is already considered "stretched" and that their response cries in children's games can greatly exceed that duration. The same is true in this corpus, with many examples exceeding 2.2 seconds (if measuring the duration of voiced production).
Let us further contrast the moan sounds with other nonlexical vocalizations reported in the interactional literature. To begin with, moans are reminiscent of pain sounds, in that they index suffering, but the sounds themselves are more similar in phonetic properties to reports of children's pain cries (Jenkins & Hepburn, 2015) or disgust responses (Wiggins, 2013), which include similar vowels and duration, whereas those of adults (Heath, 1989) are breathier or are unvoiced. Also, unlike pain sounds, the moans do not occur immediately after the game event occurs but involve a small silence, whereas pain sounds occur immediately upon triggering pain. This may be due in part to the projectability of the events; patients reporting pain may be able to anticipate that it is coming (but see Weatherall et al., 2019), whereas the events in the game are engineered to be surprises.
With respect to surprise, we can compare moans to surprised receipts (Wilkinson & Kitzinger, 2006) and to news receipts. The former study does not report specific phonetic properties (other than being "punched up," p. 154), but the transcripts show mostly notations for high pitch, which does not match the general low or falling pitch contour of moans. Studies of news receipts also report different properties than found in the moans. For example, moans begin at a lower pitch than a speaker's average, which Freese and Maynard (1998) report with bad news. However, like Couper-Kuhlen (2009), they also report that bad news receipts come with a narrow pitch contour range, whereas moans in the current data have a wide pitch contour range. Moans also involve abrupt changes at the start of production and are mostly louder than surrounding talk (features found with good news in Freese & Maynard, 1998) but have prolonged vowels (found with bad news, Freese & Maynard, 1998). Moans are closer to Oh-news receipts analyzed by Local (1996), in having utterance prominence, potentially being creaky, and being diphthongal. However, the diphthongs are not like what Local finds (e.g., not [o]->[u], likely given that they are not variants of oh-receipts) and involve other vowel possibilities. Furthermore, given that many occur in chorus, they cannot be said to occur "in the clear," as Local finds. Most of all, these studies of news receipts involve tellings of past events, rather than scenarios like games where all participants live through the event as it happens. Storytelling and playing a game are different activities; in other words, the stake for participants is not as story recipients but as coenactors of the event. Accordingly, we see different practices, both for doing different actions and for making sense of the activities as different. In looking for prior reports of nonlexical response tokens, we see that the vast majority of literature focuses on storytelling or newstelling activities (Aarsand & Aronsson, 2009 being the exception).
The final component that distinguishes moans in the current study from reported responses to tellings is that they involve nonseriousness, as discussed earlier. However, although laughter and laugh particles (Potter & Hepburn, 2010) are one of the most ubiquitous and obvious signals of potential nonseriousness, they do not occur in moans or in the lexical expansions immediately thereafter, except from other players (Extract 1, line 4; Extract 2, line 14). This is in contrast to what Jefferson (1984) finds, where laughter in troubles talk only comes from the troubles teller, not the recipient. Here, the recipients treat troubles (though not a troubles telling, in fact the recipients have often caused the trouble) as laughable. In this way, players treat the suffering as nonserious. This is not to say that players do not exhibit troubles resistance in other ways; Extract 3 shows the player producing a downgraded response to the situation in the same turn, "Oh well" (line 2). In dismissing the trouble as minor, the player closes down the relevance for other players to treat the trouble as requiring much empathy or rectification or as indexing something that would challenge the validity of the game (e.g., unfairness). Goffman (1978, p. 805) likewise predicted that a response cry "allows the recipient, or rather hearer, to treat the sound as something to which a specific spoken reply is not required." The only example involving laughter particles in a moan-like sound is not actually a moan, as it does not respond to a game event but reports on it. In the following extract, the Spider player is recounting and depicting his reaction to a game event that happened 20 minutes prior (although he did not make this reaction visible at the time). Spider explains what happened that prevented him from doing a move he wanted to do, depicting his prior reaction as "like hah::: (h)n(h)n." In telling of prior troubles, Spider follows the troubles resistance pattern described previously. The token is also embedded in a syntactic clause (for "like" as an introducer of depicted affect, see Fox & Robles, 2010); moans stand alone as turn construction units, without syntactic framing. Similar syntactic independence has been reported with prior studies of response cries and ideophones (Aarsand & Aronsson, 2009;Dingemanse, 2018;Goodwin, 1996).
Before moving to an analysis of the sequential components, compare the following two extracts, which have different nonlexical vocalizations. First, a close back vowel, namely "ooh" ([u]), occurs with other cues that cumulatively did actions like "noticing." The "ooh" response often occurs at moments where the outcome of a move was not yet certain (i.e., the potential consequences for a given player were unclear). It accomplishes receipt of yet-unfinished-game-moves. In the following, Sarah is placing a battleship as one step in her overall game move.
Extract 5: TT Star Trek Catan 38:40 (see Figure 2) Sarah's move gets several positive assessments from the other players (lines 3, 6). Emma's "oo" (Figure 2) receipts Sarah's move and treat it as noticeable, as well as demonstrating Emma's own attention to the turn's progress. However, Sarah has not yet completed her move, and the consequences of her move are about to be made clear (not shown, foreshadowed line 5). Furthermore, her move does not create a specific event of suffering for any given player; Sarah simply adds one point and a special ability. The "ooh," then, does not occur in the same sequential location as a moan (in not following a complete event), it does not accomplish the same action (in reacting to an event with interest rather than protest), and it has a different phonetic property (in the vowel quality).
Players also used nonlexical vocalizations while "thinking" about their next move (see Hofstetter, in press), which meant the sound occurred before the next game event had happened (and thus before the usual moan sequential slot). However, there was also phonetic differentiation, in that these "thinking" sounds also tended to lose their voicing comparatively early in production. For example, in the following, it is Spider's turn to play and has been for several minutes. He has repeatedly delayed making a move.
Extract 6: 160908 Dominant Species 2_0:2:50 (see Figure 3) Although Spider projects (not for the first time) that he will play imminently, he hovers his arm and piece above the board, delaying his move (lines 3-8). As he continues to delay, he narrates that his activity is "so difficult" (line 7), prefaced with a click and a nonlexical sound. The vowel quality is similar to a moan, but much of the production time is actually an unvoiced exhalation ( Figure 3); the prolongation of the vowel is abandoned early, while the voiceless exhalation continues. Thus, "thinking" sounds have one potential difference in phonetic production (duration of voicing) and occur at different sequential moments in the game. In summary, moan sounds involve a cluster of phonetic properties: prolonged production of a set of vowels (combinations of [ɐ], [ɑ], and [o̞ ] most often), typically involving a pitch contour with a wide span. The moans do not resemble previous reports of nonlexical responses in either phonetics or activity and also are not treated as serious displays of trouble. They are systematically treated differently than other sounds produced in the data corpus such as prolonged [u] or sounds where the voiced portion is short through produced with a prolonged voiceless sound. The next section will focus in more detail on the sequential contingencies of moans.

Sequential consequences of moans
In this section, I will analyze what actions follow the production of moans. After the production of the moan token, the suffering player may downgrade their response, which is more congruent with findings on everyday management of troubles (e.g., Jefferson, 1984). Goodwin (1996, p. 394) also reports a similar format, in which a triggered event prompts a response cry and some subsequent elaboration. As Goodwin writes, "the response cry thus locates (or at least notes the existence of) some other event and formulates in a particular way, i.e. as something that has the power to elicit the strong reaction visible in the cry." Wiggins (2002, p. 327) also finds the pattern of (food enjoyment) vocalizations being followed with lexical evaluations and points out that the reverse sequence (lexical followed by nonlexical) can appear contrived or inauthentic. Reactions to the game event must therefore occur promptly. Furthermore, the moment after a game event is also a slot for other game work to occur (taking the next game turn, clearing up that round, etc.), so there are competing actions imminent. As seen in Extract 2, the reaction slot provides opportunities for players to display joint attention and shared reactions to the events. A particularly effective way to demonstrate that a reaction is shared is to produce it chorally (see Lerner, 2002). For example, in the following extract the game is a speed game. John is the first to finish his set of cards and announces "Blitz" (line 3), which ends the round and prevents the other players from finishing their own set of cards. The other three moan in choral response (lines 7-9).
Extract 7: 180815 Dutch Blitz_12.58 (see Figures 4 and 5) A moment after John announces he has ended the round, and during which the players look at the table and cards, the players all moan together (lines 7-9). The joint production of the moan takes work to accomplish, however. Kat begins a moan sound that is abandoned (line 6), as no one joins her moan (which may indicate she has misread the situation and must reassess). Kat had timed her moan to land in beat with a rhythm set up by John's card plays (one pretranscript, one line 4); this timing (0.8 s) can be seen in Figure 4. Ogden (2013) found that clicks can be used to set up rhythms that then do projection work, and the card placement sounds appear to do the same here. Gwen and Adam begin their moans delayed, relative to that rhythm. Kat rejoins the moan on a half beat, setting up a rhythm a half beat later, which Gwen continues in concluding her moan with a click on that new beat/off-beat (from the original rhythm). The vowel is also coordinated, with Gwen and Adam picking up similar vowels from Kat's initial cut off moan. Finally, the pitch contour is also coordinated, especially between Kat and Gwen who (in the second moan for Kat) lower their pitch at the same rate, maintaining a single contour (see Figure 5). The quieter voice, Adam, drops lower at the early portion of his moan (see Figure 5). The players coordinate the features of the moan and display a shared experience, which matches their shared fate in the game event.
The completeness of a game event is not always straightforward. Sometimes players "try out" game moves before redacting them and trying again, so it is not always clear if a move is certainly proceeding. Furthermore, giving a moan too early may give away whether the move would cause suffering for a competitor, and players have a stake in hiding which choice is the most impactful until the last minute. Players will delay producing a moan until the event is certain to have occurred and certain to have caused suffering of some kind. The next two extracts each show a move unfolding and ending up in a moan. Extract 8 shows a player (Kat) working out how the game move is supposed to function, and this uncertainty leads to clarification before the moan. Extract 9 shows a player (Jasmine) slowly undertaking a move against a competitor, and this dramatic slowness makes the upcoming damage projectable. Upon the moves being finalized (lines 13 and L8 respectively), the suffering recipient moans.  1 and 2). These may account for Kat's "early" onset (it is timed with the beat), and her rejoining the moan at a half beat later. Gwen completes her moan on the half/off beat with a click. As Kat works through her game action, she "kills" one of John's pieces. John receipts this action with an "ooh" sound and a confirmation request. Kat has shown some uncertainty about what her chosen move entails (lines 1-2), so John's confirmation request may be an effort to be more certain and/or a way to invite Kat to do a different move that harms John less. When Kat's move is confirmed, John's moan follows. Kat reconfirms with a more sympathetic utterance before initiating the next activity with "uhm." In Extract 9, Jasmine projects the upcoming event both verbally (line 4) and by a very slow movement of the "wessel" token toward Cal's tile (the slowness is likely for the sake of drama, rather than solely uncertainty). Cal can thus project where she is going to put it and that he will suffer because of her putting it there. Just after the moment of placement, Cal transitions from a lexical cry, "no," which indexes denial, to the nonlexical moan, which does not deny the fact of Jasmine placing the vessel there. The players in this extract are less sympathetic, with Sarah encouraging Jasmine to "bury" Cal (line 9). Across these two extracts we see the players orient to the certainty and completeness of the game event, timing their moan cry to occur after the event has been made certain (although also ratifying the event as complete with their moan).
Extract 10 also shows a way in which players downgrade the severity of suffering. This is a common follow-up to a moan sound. Typically, the downgrade comes from the sufferer themselves, as in Extracts 1 and 3 earlier and Extracts 11 and 13 (for more on downgrades from the inflictor of suffering, see Extracts 12,14,15). For example, in the next extract, Alan needs a specific number on his dice roll (they are special dice, with some blank faces), which he fails to get. His first reaction is a moan (line 8), but his second reaction is a comparatively minimal assessment of bad luck, "bummer" (line 11).
Extract 10: GN Small World_39:20 In response to the dice roll, which is not sufficient for Alan's game move, Alan moans (along with Harry, line 9, while Tina cheers, line 10). His second response, "<Bummer.>" (L11) marks a significant contrast to the moan. "Bummer" is an assessment of the suffering experienced as mild, which does not show the affectual involvement seen in response cries (as reported by Goffman, 1978;Reber, 2012; but also seen in orientations to moans in previous extracts). In terms of prosody, the "bummer" utterance is quieter and in Alan's normal pitch range. It is also followed by an activity shift marker, "okay," which displays that Alan is not lingering on the suffering moment but progressing the game along right away. This quick shift stands in contrast to troubles tellings where there is a longer stepwise exit (Jefferson, 1984). It is more similar to the exit of nonserious sequences like teasing that have been reported, wherein participants preface their turns with "no" and switch from giving nonserious responses to serious ones (Drew, 1998;Schegloff, 2001). However, a "no" preface here might be inappropriate as the player is not reversing or rejecting a prior stance but rather downgrading it.
Another example of the contrast between the moan and subsequent assessment is available in the next extract. Hal is facing a series of monster cards. If they inflict too much damage on him, he will die and lose the round. He sees the damage upon flipping the card, so there is suspense with each flip whether he will survive.
Extract 11: GN Welcome to the dungeon_0:34:10 When Hal flips the card, it is a monster that damages him enough to kill him. Larry and Tina react first. Hal gives a brief moan (line 9), but it is not until he confirms with the others that he has definitely died that he gives a longer, fuller moan sound (line 13). The players discuss how close Hal was to winning. Hal's continued laments that he was very close could be treated as troubles or as complaints. However, Hal concludes by describing the situation as "awesome" (line 22), downgrading its severity-in fact, refocusing entirely from the suffering event to the exciting scenario that the game created. The "That's awesome" utterance is prosodically different than the moan, being quickly produced in Hal's normal pitch range, and quieter than the moan. Both Hal and Stephanie treat this final assessment as transition relevant, with Hal issuing an activity shift marker and Stephanie prompting the next player (line 26). Ogden (2010) has also found sequence closing complaints to precede an activity shift and occur with likewise quieter, lower production (although the complaints here are summarized in with a noncomplaint).
When players do not downgrade, others may do it on their behalf. However, this involves a different distribution of participant rights, which may make it more difficult to maintain the nonserious element. For example, in the following, Beth steals coins from Sally, who moans (line 6) and protests she would not have harmed Beth with her money (line 8). Beth downgrades the severity of stealing "only two" coins (line 12).
Extract 12: RPS 170404 Coup_0:29:50 Beth downgrades the seriousness of the suffering for Sally (line 12), rather than waiting for Sally to do it for herself or ignoring the lack of a downgrade. Note that this is not the same as rectifying the event, as Beth does not return the coins or change her game move. Beth seems to orient to the inappositeness of her downgrade immediately afterward. First, she suggests Sally kill Jim (instead of herself, Beth). This is a potentially nonserious suggestion, given that Sally can no longer kill anyone on her next turn, due to lack of coins, and due to the likelihood of retribution. Second, Beth appends postcompletion laughter particles (Shaw, Hepburn, & Potter, 2013), which is often done in moments where participants manage potentially problematic prior action and as a means of managing complaints (Holt, 2012). Sally, however, crosses her arms in a quite abrupt, showy fashion, possibly indexing a mock display of annoyance or pouting. In this extract we see the potential for difficulty when downgrading on someone else's behalf (see Extract 15).
The practice of downgrading on someone's behalf may overstep rights, but empathetic expressions do not do so. Both Extracts 10 and 11 show the players who are not suffering the event producing moan sounds, suggesting an expression of empathy with the sufferer. Extract 13 shows another instance. Tina is rolling the dice and needs a three, but she does not succeed.
Extract 13: GN Small World_0:56:15 The other two players also produce moans, in choral overlap with Tina (lines 9-11). They each acknowledge how close Tina was to success (she rolled a two), with Harry additionally acknowledging how "fun" a successful roll would have been (line 12). Their responses both empathize with Tina's fate as well as ratify the failure. Tina produces a contrasting, minimal complaint "boo" (L13), before shifting activity to the next stage of her game turn. Such "moans from nonsufferers" occur in both competitive and collaborative games (see also Extracts 9 and 10) and not only during moments when players experience a common fate (as in Extract 6). In the previous extract, Harry and Alan both benefit from Tina's loss, in terms of getting closer to winning. Their stake in Tina's failure does not prevent them from publicly indexing affiliation with Tina's situation. Moans, then, are not solely expressed in response to one's own suffering but also to others.
In summary, then, moans are followed by contrasting downgrades of the severity of the game event. While moans may be treated as displays of (suffering) affect, the players regularly follow up with displays of troubles resistance that treat the event as expectable, exciting, interesting, or otherwise not severe. This troubles resistance is a display of "sportingness" (Hofstetter & Robles, 2019) or willingness to experience the ups and downs inherent in games and play. The sequence of moan-then-sporting-willingness accomplishes the same shift as Drew's (1998) "po-faced," serious responses to teases, and Schegloff's (2001) "no"-prefaced follow-ups to jokes. Players make relevant both the affect-laden reaction and the "calmer" reaction. The contrast between them highlights the nonserious nature of the game.
Next, I will compare moans with lexical responses to game events that cause suffering.

Lexical responses
Moans are nonlexical, in that they do not have a conventionalized form-to-meaning relationship (or, at least, less conventional than most words). In addition to advantages such as displaying spontaneity and immediacy (Wiggins, 2002), and being easily adopted for choral responses (Lerner, 2002), moans also have the advantage of not indexing a specific, conventionalized word. For instance, "no" is a possible response to game events, but it inevitably also indexes denial. Extract 9 showed an example of a player switching from just such a denial to a nonlexical moan upon the event being certain and thus ratifying the event as valid, which a denial might fail to achieve.
However, players sometimes do respond to events with lexical tokens instead of moans. This causes the conventional lexical associations to be made relevant, bringing their own contingencies and accountabilities. In the next extract, Mammal is killing one of Spider's pieces. He reacts with a denial, produced in a high-pitched cutesy voice (line 3). However, Mammal responds seriously by complaining about Spider's prior conduct.
Extract 14: 160908 Dominant Species 1_0:51:10 Spider complains in a high-pitched voice with prolonged syllables, sounding child-like and potentially nonserious. Mammal takes Spider's question seriously, giving an answer in her normal tone of voice (line 8). Spider's denial and question also lack a downgrade of his suffering (line 7 deals with where he ought to put the killed piece). Spider smiles at Mammal as he finishes putting the killed piece away, but she does not reciprocate. Thus, Spider's attempts at nonseriousness are much less successful than the moans described earlier.
In the final extract, Reptile is picking between two cards. In picking a card, he uses it up, and it is no longer available for other players to use. When Reptile chooses, he picks the card that Spider was hoping to use, and Spider complains about this event. This sequence includes one of the few apologies in the corpus (line 10). Spider's reaction is a lexical utterance, a complaint. Furthermore, Spider upgrades the complaint (line 7) from its first version to claim that Reptile has done something to Spider, making Reptile personally accountable. The moans seen throughout this article rarely made accountability relevant. Furthermore, Spider here upgrades his complaint, rather than the downgrades usually seen after moans. Spider pursues his complaint past the apology, with mock crying (line15) and explanations of the damage that Reptile has caused (lines 20, 25-26, omitted lines). Mammal can be seen to treat Spider's talk as inapposite, hindering the progressivity of the game, as she attempts to initiate the next activity multiple times (lines 19, 28). Bird eventually sanctions Spider's talk as pointless lamenting (line 52) and explicitly says Spider no longer has the options he is describing. Bird, who has no stake in the matter, is better positioned to sanction Spider's conduct as inapposite than Reptile.
Both Extract 14 and 15 demonstrate that lexical responses are more vulnerable to interpretation as being serious and/or as being complaints with a blamed aggressor (rather than "mere" expressions of suffering). Moans do not typically have the same vulnerabilities. The contrast here demonstrates the reliance on "vagueness" that has often been reported with respect to nonlexical vocalizations (Reber, 2012;Wiggins, 2002).

Discussion
In this article, I have described the nonlexical vocalization that I call a "moan", which I analyzed according to their use in interaction rather than as signifiers of emotional content (consistent with, e.g., Kockelman, 2003;Reber, 2012). It is a sound produced after a game event that receipts the event as complete and valid and is treated as an affectual expression of the player's reactionnamely, an expression of suffering. It is often followed by a downgraded, lexical utterance that restates the reaction in a way that displays resistance to trouble or willingness to proceed ("sportingness," Hofstetter & Robles, 2019). Moans involve a cluster of phonetic propertiesnamely, a prolonged vowel with a wide-pitch span contour, with open vowels, typically open back vowels but sometimes open mid or central vowels. Moreover, moans never involve close vowels ([u]) nor voiceless production, both of which are treated as accomplishing other actions (see Extracts 5 and 7). Moans avoid certain lexical contingencies; for instance, response words like no have the inevitable consequence of indexing denial or "halting" (see Keevallik, 2012), which may not be appropriate for moments when they are ratifying a move as complete (see Extract 9). Finally, moans have a role in coordinating game-based turn taking by simultaneously receipting a move as valid and demonstrating ongoing, "genuine" engagement with the game activity.
Moans achieve their work in several ways. First, the sequential placement constructs a sense of immediacy and spontaneity, which promotes the sense of the reaction being authentic (Wiggins, 2002). Second, moans set up contrasts between the initial, "flooding over" response and the follow-up, lexical downgrade. One contrast is between the affective valence; moans are treated as an expression of affect like suffering, disappointment, anger, etc.; whereas the follow-up utterance is treated as an expression of troubles resistance. The other contrast is phonetic; moans are loud, nonlexical, and prolonged, among many other indicators that often come with "heightened" affect (Selting, 2010, p. 238); whereas the follow-up is quieter, lexical, with less emphasis, no prolongation of syllables, etc. (and where that is not the case, it tends to co-occur with problematic follow-ups, e.g., Extract 12). Moans also contrast by virtue of using sounds that are unusual in English; not many native English speakers may realize they are perfectly competent users of epiglottal trills.
The previous contrasts are interactional manifestations of Bateson's paradox. Bateson said that the message of "this is play" is a paradox-specifically, "a negative statement containing an implicit negative metastatement," i.e., "These actions, in which we now engage, do not denote what would be denoted by those actions which these actions denote" (1972/2006, p. 317). In interactional terms, the sequential implications are not what they would otherwise be (Glenn, 2003). Intensity (in moans) followed by relative mildness (in downgrades and continuing play) designs the turns to demonstrate those paradoxical, nonserious implications. Thus moans, and these contrasts, show us another way to do nonseriousness than laughter.
Moans achieve their work without players necessarily attributing any specific emotion or affect. Moans function without players needing to know whether a moan indexes disappointment, frustration, anger, a combination of these, or indeed, a bluff (players do not need to "bin" emotions, see Sidnell & Enfield, 2014). There is no need for the player to be displaying some particular internal experience (even if one is occurring) in order to be accountable for the moan as a local receipt of the game event (see also Kockelman, 2003).
Finally, mirroring the imprecision with affect, moans are in an analytic "grey zone" concerning conventionalization: They are precise in some elements, such as sequence, and imprecise in others, such as vowel quality. This presents no problem to participants; instead we should critique the binary conceptualization of conventionalization. It is possible that moans are becoming conventionalized, but the players are making sense of the "partially" conventionalized token now. The degree of conventionality should be treated as a members' problem; members make sense of the vocalizations, but they also use the changeable form in playful ways. Indeed, as moans rely on being treated as spontaneous, "flooding out" sounds, they may require a low degree of conventionalization in production to achieve authenticity, as well as to achieve the contrast with lexical assessments of the event-in essence, "doing" unconventionality. This suggests that nonlexical vocalizations may be integral to and inevitable in a complete language system, even if previously considered marginal (Dingemanse, 2018).

Funding
Thank you to Alex Stein for financial support during data collection and fieldwork. This article was supported by a grant from the Vetenskapsrådet [2016-00827].