Introduction
Everyday interaction features thousands of sounds such as um, uh, oops, ouch, meh, tch, yum and ugh. A growing body of research shows that these sounds are applied in ways that are interactionally and socially meaningful (e.g., Wright, 2011; Ogden, 2013; Hoey, 2014, 2020; Culpeper and Oliver, 2020). In the current study we follow Reber (2012) and Reber and Couper-Kuhlen (2010) in referring to them as “sound objects” (Reber and Couper-Kuhlen, 2010; Reber, 2012), although they are also referred to by various other terms including “non-lexical vocalizations” (e.g., Keevallik, 2021), “vocal gestures” (following Mead, 1972; Dietrich et al., 2007; Harkness, 2011), and “pragmatic noise” (Culpeper and Kytö, 2010; Culpeper and Oliver, 2020).
In this paper, we explore the potential politeness-related functions of a hissing-like sound object that occurs in Korean. This hissing sound, which was first reported for Korean in Winter and Grawunder (2012), is a lateral ingressive slick produced by drawing air along the sides of the tongue or between the teeth. Although it does not appear in dictionaries, Korean speakers may transcribe it as 씁 ssup. We will refer to this sound as a “hiss,” even though “hissing” is generally used to refer to egressive sounds produced with air coming out of the lungs, rather than being sucked in as in the case of 씁 ssup. The reason for calling this sound object a “hiss” is that it is a very loud and salient high-frequency noise that one would auditorily describe as having a “hissing” quality. Moreover, we use the term “hiss” because past research used the same term for this particular Korean sound object (Winter and Grawunder, 2012; Kim et al., 2021).
Winter and Grawunder (2012) followed by Kim et al. (2021) both showed that speakers use this hissing sound more frequently when addressing a status superior than when addressing a status-equal friend, hinting at a politeness-related function given that indexing social distance is an important politeness practice in Korean performed primarily by grammaticized honorifics, but also by distinct acoustic and gestural behaviors (see Section Breathing and Hisses in Korean and Beyond below). However, neither paper investigated this matter in any depth, or establish any detailed reasons why hisses might be used more frequently toward a superior. We set out to fill this gap in the current paper.
Our goals in this paper are to establish the pragmatic functions of Korean hisses and use this analysis to explore how sound objects can take on politeness functions. To do this, we revisit the datasets used in Winter and Grawunder (2012) and Kim et al. (2021) and add one further dataset (Brown and Winter, 2019). By bringing the study of sound objects into the remit of politeness research we aim to contribute to a growing body of research that sees politeness as a multimodal phenomenon that is manifested across a range of modalities in addition to verbal speech (Brown and Prieto, 2017).
Background
Multimodal Politeness
“(Im)politeness” involves the mediation of interpersonal relations via a range of different behaviors, including conventionalized acts of etiquette and rudeness, and the ritualized marking of formality and hierarchy. According to Brown P. (2015, p. 11,620) politeness is “a matter of taking into account the feelings of others as to how they should be treated, including behaving in a manner that demonstrates appropriate concern for interactors' social status and their social relationship.”
Among the many different sub-categories of politeness, in this paper we focus on ways in which speakers modify their language according to social distance. We follow Leech (2016, p. 126) in seeing social distance as including both “vertical distance” (i.e. distance in terms of power or authority) and “horizontal distance” (i.e., distance in terms of solidarity, familiarity, and intimacy). Research has shown that social distance is one driving factor of politeness-related expressions (see Goldsmith, 2007, p. 227 for overview). Holtgraves and Yang (1992) found that Korean and American respondents used more polite request strategies when addressing superiors and strangers, with Koreans varying their responses more strongly. This mode of politeness has also been referred to in the literature as “deference” (Brown and Levinson, 1987), “discernment politeness” (Ide, 1989), and “bivalent politeness” (Leech, 2014).
(Im)politeness is communicated not just verbally, but also multimodally via other communication channels. This is evident from the fact that the politeness level of an utterance can be completely altered by nonverbal context. For instance, Culpeper (2011a, p. 57) points out that the seemingly “polite” question “Do you know anything about yo-yos?” can be rendered impolite by heavily stressing the beginning of “anything” and delivering the remainder of the utterance with sharply falling intonation.
On the acoustic level, politeness-related meanings are communicated via a range of cues, including pitch and intensity (loudness). Universalist approaches have claimed that high pitch is associated with politeness in all cultures (Brown and Levinson, 1987; Ohala, 1994) because high pitch signals small size and can therefore indirectly also signal submissiveness. However, a recent meta-analysis of data from seven languages (Winter et al., 2021) found that pitch is either lowered (Korean, Russian, Catalan, German, Austrian German), or does not differ (Japanese and Chinese) when using polite language toward a socially distant superior. In Korean, speakers seem to prefer a low, monotonous and quiet voice when politeness is required (Winter and Grawunder, 2012). In addition, Korean speakers, especially female speakers, hyperarticulate vowels when speaking politely (Oh et al., 2021).
Politeness-related meanings are also marked by the ways that speakers perform gesture and other bodily movements. Overall, body movements become more curtailed when politeness is required: they are less frequent, smaller, and less animated (Brown, forthcoming; Brown and Winter, 2019; Hübsch, under revision). The types of gestures that accompany (im)polite language may also be influenced by culture-specific associations between conventionalized gestures and (im)politeness. For instance, in Yoruba, index finger pointing toward a referent who is older may be impolite, although open-hand pointing is acceptable (Orie, 2009). In the current paper, we extend the study of multimodal aspects of politeness to one further domain: sound objects.
Sound Objects
“Sound objects” refer to various noises that speakers make during interaction, including fillers such as um and uh, exclamatory sounds (oops!, ouch!, yum!), and various clicks, whistles, sighs, sniffs, and breathing-related noises. Until recently sound objects were not widely researched, perhaps due to the assumption that they were “the flotsam and jetsam of mere performance” (Dingemanse, 2020, p. 188). In fact, sound objects may in some instances have advantages over lexicalized vocal expressions, such as for expressing affect in more immediate and genuine ways (e.g., saying mmm instead of saying “it tastes good”) (Wiggins, 2002), and expressing abstract qualities such as movement, shape and timing, for example, during dance classes (Keevallik, 2021). Sound objects are used systematically for a wide range of functions, and their occurrence is sequenced and synchronized with verbal speech, as well as prosody and gesture.
Sound objects play important roles in the organizational sequence of interactions. Filler noises such as um and uh signal to the hearer that the speaker is busy planning their turn and should be given an opportunity to finish (Tottie, 2011, p. 403), but can also be used to invite the interlocutor to complete the utterance (Clark and Fox Tree, 2002). Turn completion in English can also be marked by sniffing (Hoey, 2020) and sighing (Hoey, 2014), whereas speakers of English (Wright, 2011; Ogden, 2013) and Mandarin (Li, 2020) use clicks to mark the onset of new and disjunctive sequences of talk. Speakers can also use sniffs (Hoey, 2020) as well as swallowing (Ogden, 2021) to delay the progression of talk. Wiggins (2002, 2013) demonstrates that nonverbal vocalizations indicating gustatory pleasure and disgust (such as mmm and eugh) invite collaborative assessments of food and also table manners.
Sound objects can also communicate a range of different affective and social stances. Clicks are used to signal disapproval through practices conventionalized as “tutting” (Ogden, 2013, p. 300), whereas post-completion sighs not only mark turn completion but also indicate a resigned stance toward the talk. In Egyptian Arabic, grunting communicates various emotional states, such as feeling disgusted or fed up (El-Bahrawy, 2007). In Caribbean Creoles, speakers use “kiss-teeth” (a noise produced “by an ingressive airstream captured in an air and saliva pocket created in the mouth through varying configurations of velar, dental and lip closures,” (Figueroa, 2005, p. 74) to negotiate moral positioning, community norms and politeness. Finally, Korean uses “fricative voice gestures” (FVGs)—rough, voiceless sounds similar to the sound made when clearing the throat—to signal personally felt intensity (Harkness, 2011; Winter et al., 2019).
The use of sound objects can be a locus for sociolinguistic variation. Younger speakers have a higher um:uh ratio (Tottie, 2011; Laserna et al., 2014) in English and other Germanic languages, with women being a generation ahead of men in this change (Wieling et al., 2016). In Korean, whereas male speakers emit a khu noise after downing a shot of liquor, female speakers may prefer to emit khya, since the latter is associated with softness and femininity via the substitution of the high back vowel [ꟺ] for an open low-back vowel [a] (Harkness, 2011; Winter et al., 2019).
Sound objects frequently appear in synchrony with gestures and other body movements. For instance, Ogden (2013, pp. 311–314) shows that clicks often occur with swallowing, lip smacks, and removal of gaze. Caregivers comforting children in preschools in Sweden synchronize elongated, soft and whispery sh::: noises with physical soothing responses such as stroking and embracing (Cekaite and Kvist Holm, 2017). When feeding their infants, English-speaking parents produce lip smacks to encourage and endorse eating. These lip smacks co-occur with stable gaze on the infant, smiles, raised eyebrows, and head tilts (Wiggins and Keevallik, 2021). Sound objects thus work in unison with other multimodal cues to signal various interactional stances, as we shall see for politeness in this paper.
Breathing and Hisses in Korean and Beyond
The current paper looks at a type of nonverbal speech sound that we will refer to as “hissing.” This sound object has not yet been studied in depth in previous research. It is produced by rapidly inhaling with the lips partially opened. Winter and Grawunder (2012) describe this ingressive sound as sounding similar to a lateral fricative (p. 810), and it can be produced with noisy frication arising from the sides of the tongues. Alternatively, these hisses can also be realized as a bidental fricative, with frication arising from in between the teeth. Such ingressive hissing sounds have rarely been mentioned in the literature on other languages, although similar phenomena may exist in Danish (Fredsted, 2005, p. 169), Chinese (Key, 1975; Eklund, 2008), Japanese (Critchley, 1939), and English (Brooke, 1988).
The intake and outtake of breath is known to play an important role in the organization of talk. Given that air needs to be breathed in first to produce speech, inbreaths can work as a cue for speech initiation (Jefferson, 1986; Schegloff, 1996). Speakers tend to produce more audible breath intakes before turn initiation to signal that they are about to take the floor (Ishii et al., 2014; Aare et al., 2015). Breath holding can be used as a turn keeping device, whereas releasing breath works to yield the turn to the next speaker (French and Local, 1983).
Speakers have the ability to manipulate the placings and design of their breath intakes and outtakes (Hoey, 2020, p. 119). For instance, inbreaths can be realized as gasps that indicate realization or shock (Lerner and Linton, 2004) or suck-teeth, which marks meanings such as disgust, defiance and disapproval in African American English (Alim, 2004). Ingressive hissing can also be used to express felt or anticipated pain (Diller, 1980). Meanwhile, outbreaths can be stylized as sighs of resignation (Hoey, 2014) or, in French as pf sounds that mark speaker disengagement (Baldauf-Quilliatre, 2016).
The hisses that are the focus of the current investigation can be thought of as a special form of stylized breath intakes, with the frication and resulting high-frequency noise making them very salient. For Korean, Winter and Grawunder (2012) and Kim et al. (2021) found that hisses appeared more frequently in speech addressed to status superiors, and less frequently in speech addressed to intimates. Winter and Grawunder (2012, p. 813) note that hissing might occur more frequently in formal speech due to the increased effort that it takes to produce noisy breath intakes, with the Effort Code (Gussenhoven, 2004) predicting that polite language tends to be more effortful. Moreover, they propose that similar to how ah and oh can sometimes function in English to make the process of planning sentences audible to interlocutors (Clark and Fox Tree, 2002), the hisses could serve the function of signaling that sentences are planned with care by filling otherwise inaudible pauses with clearly audible phonetic material. Meanwhile, Lee and Lee (2021) found that Korean hisses signals an epistemically “non-committal” stance; however, their research only looked at hisses that occurred in turn-initial responses to information-seeking questions. The current study addresses the need to look more deeply into the functions of hisses, and to establish the reasons why they might appear in speech addressed to a status superior.
Methodology
Data
This paper re-analyzes the datasets used in the two previous studies that reported that speakers use hisses more frequently when addressing a superior: (1) an oral discourse completion task from Winter and Grawunder (2012), and (2) a corpus of dyad recordings from Kim et al. (2021). In addition, we investigate hisses in (3) a corpus of TV drama interactions previously studied by Brown and Winter (2019).
Winter and Grawunder's (2012) oral discourse completion task (DCT) involved 16 Korean speakers (9 female, 7 male; aged 21–31) producing oral responses to written scenarios in a laboratory recording environment. The first four scenarios involved formulating a spoken message that they had to leave on an imaginary cell phone voicemail (e.g., leave a voicemail giving directions to a vacation destination). For the remaining 10 scenarios, they had to formulate the first utterance of a role-play based on a given situation (e.g., apologize to your professor for arriving late for an appointment). Whereas the assumed interlocutor for half of the items was a status superior (such as a professor), the interlocutor for the other half was an intimate status equal (such as a friend). The stimuli materials, including visual images of the assumed interlocutors, were all presented on a computer screen in a sound-attenuated booth, and the utterances were recorded with a head-set microphone AKG C420 (linear characteristic) with 48 kHz/16 bit sampling. The total length of all the recordings was 1 h and 30 min. A list of the scenarios can be found in Winter and Grawunder (2012).
The dyad recordings by Kim et al. (2021) featured 14 Korean speakers (7 male, 7 female; aged 19–27) participating in two face-to-face interactions: one interaction with a friend, and one interaction with a professor. The friends were actual same-gender friends of the participants, whereas the professor was a confederate who was a 60-year-old male professor of English literature. The two recordings occurred on different days and were counterbalanced so that half of the participants interacted with the professor first, and half of them interacted with the friend first. During the recordings, participants performed four interactive tasks: a natural conversation where they prompted to talk about a movie they had seen recently, a retelling of a “Tweety Bird” cartoon (e.g., McNeill and Duncan, 2000; Kita and Özyürek, 2003), a map task (Anderson et al., 1991), and an apology role-play previously used in Brown (2011). The interactions were video recorded using a Canon VIXIA HFM 41 video camera. The total length of all the recordings was 5 h and 8 min.
The corpus of TV drama interactions consists of just under 3 h (2:58:28) of video clips collected from three South Korean televised drama series: Bad Guy (SBS, 2010), Pinocchio (SBS, 2014), and Two Outs in the Ninth Inning (MBC, 2007). These are all fictional mini-series that focus on the lives of young professionals. We analyzed the speech of six prominent characters from these dramas, three females and three males (one male and one female from each series), all of whom are young professionals in their mid-twenties or early thirties. For each of these six characters, we identified a same-sex workplace superior and a same-sex intimate status equal with whom they had regular interactions.1 We then extracted video clips of all dyadic interactions between these characters.
There are some important differences between each of these datasets. Only the dyad recordings are both unscripted and interactional. The oral DCT is unscripted, but not interactional (since the interlocutor was only imagined), and the drama interactions are interactional but scripted. Despite (or perhaps due to) their scripted nature, the drama interactions provide some very vivid examples of how hisses can be used alongside facial expressions and other body movements to communicate various discourse functions. The drama data also allows us to assess the extent to which hissing is reflected in Korean popular culture, which has become an important context for global access to the Korean language (see Min, 2021).
The advantage of considering all three of these data sources for the current study is that they all contain examples of the same speakers producing language in two conditions: (1) a socially distant condition where they are addressing a status superior, and (2) a socially close condition where they are addressing a status equal. Our previous analyses of these datasets show that the speakers change their behaviors in various ways between these two conditions. When addressing the superior, they used more honorifics and backchannels (Kim et al., 2021), spoke with a lower-pitched and more monotonous prosody (Winter and Grawunder, 2012), moved less and adopted more compact body positions (Brown and Winter, 2019), and used smaller and less expressive gestures (Brown et al., forthcoming). We now look in more detail at the different frequencies of hisses.
Analysis
We coded all hisses that appeared in the transcripts of the three datasets. To distinguish hisses from simple breathing noises, we followed the criterion that the hisses needed to have a clear sibilant-like quality. Following native speaker intuition regarding how these hisses should be transcribed, we represented them in our transcripts as 씁 ssup. The coding resulted in the identification of a total of 291 hisses: 102 hisses in the oral DCT, 161 in the dyad recordings, and 28 in the TV dramas.
The analysis followed two steps. We first of all coded the hisses according to whether they occurred in interactions with superiors or with inferiors in order to confirm the findings of Winter and Grawunder (2012) and Kim et al. (2021) that these noises do indeed occur more frequently when interacting with superiors.
Next, we adopted an integrative pragmatics framework (Culpeper and Haugh, 2014) to establish the social actions that the hisses are performing. Integrative pragmatics is an approach that focuses on establishing what form-function relationships mean to language users during interaction, and in particular situated, sequential contexts (Culpeper and Haugh, 2014, p. 266). This approach draws on an interactional linguistics (Couper-Kuhlen and Selting, 2017), particularly ethnomethodological conversation analysis and work on multimodality in human interaction. We analyzed the social actions performed by hisses by building collections of recurrent co-occurrences between hisses and particular pragmatic contexts, and analyzing in detail how sequential position and turn design lead to action formation (Qiu et al., 2021). Going beyond conversation analysis (which traditionally is not a field that concerns itself with politeness, Culpeper, 2011b), the integrative framework allows us to look at how these social actions have the potential for generating politeness-related social meanings.
The identification of social actions performed by hisses occurred in an inductive, bottom-up fashion. We began by tagging specific types of sequences, speech acts and linguistic patterns that co-occurred with hisses. As we reviewed more data, we grouped the codes into larger concepts until we arrived at four categories. Of these four, we noticed that the first three corresponded closely with three categories used in previous studies on sound objects (including Wright, 2011; Hoey, 2020).2 We therefore adopted terminology from these studies for convenience when naming the categories, although it should be emphasized that this was done in a bottom-up fashion as the categories emerged, rather than trying to match the hisses to pre-existing categories.
Displaying delay and delicacy: The speaker hisses when attempting to delay the initiation or progression of a turn at a delicate point in the interaction, namely before a dispreferred response.
Activity shift: The speaker hisses before transitioning to a new sequence of talk.
Word search: The speaker hisses when attempting to recall a word or detail.
Conveying skepticism: The speaker hisses before expressing a questioning attitude or doubt regarding the truth value of the prior turn made by the interlocutor.
The entire dataset was double-coded by two independent coders, who were the first and second authors of the papers. The second author is a native speaker of Korean, whereas the first author is a second-language speaker who uses Korean at home with his family. Initial levels of agreement exceeded 90% and discrepancies were discussed until consensus was reached and each token was distributed to one “best fit” category. We view these four categories as being inter-connected rather than discrete groupings. In our presentation of the data, we discuss cases where tokens displayed characteristics of more than one category.
In line with moves within pragmatics and conversation analysis to view interaction as a multimodal achievement, we analyzed bodily visual practices (Ford et al., 2012) (i.e., gaze, head movements, gestures, etc.) that occur when hisses are performed, although this was only possible for video-recorded data (dyad recordings and drama interactions).
Raw counts of the hisses are available under the following repository: https://osf.io/8ud7q/.
Discussion
The analysis has confirmed that hisses do indeed occur more frequently when speakers are addressing status superiors, and has demonstrated through a close qualitative analysis that hisses perform four distinct types of social actions: displaying delay and delicacy, activity shift, word search, and conveying skepticism. The higher frequency of hissing with superiors was attested in two of the three datasets, and all of the social actions except for conveying skepticism were present across all three types of data, confirming the robustness of the findings.
Generally speaking, hisses are occurring directly preceding or during some kind of relational or interactional trouble spot. This trouble spot may be rapport-threatening (as in delay and delicacy hisses, and also skepticism) or conversationally disruptive (as in activity shift or word searches). In these contexts, hisses stop the progression of talk, while allowing the hearer to initiate or maintain their turn. The hisses are often accompanied by other devices that delay talk progression, including uh/um-type fillers (examples 1 and 5), unfilled pauses (examples 4 and 7), audible breath intakes (example 4) and lexical repetitions (examples 3 and 8). They are also often synchronized with bodily visual practices such as head tilts (example 10) and gaze avoidance (examples 3, 4, 8, 9) that mark memory retrieval, but which in these sequences may become stylized as markers of uncertainty and delay. We thus see the underlying core meaning of hisses as involving difficulty, uncertainty and hesitation.
By drawing in breath in an audibly loud way, speakers create the appearance that they are going to say something, given that inbreaths precede speech and that louder breath intakes are cues for speech initiation (Jefferson, 1986; Schegloff, 1996; Ishii et al., 2014; Aare et al., 2015). But by only hissing rather than actually saying something concrete, they may also mark their reluctance to perform problematic talk. This display of reluctance can take on politeness functions, particularly in the case of sensitive speech acts, since it shows that the speaker is hesitant to impinge on the hearer. Indeed, this politeness-related function of hesitation is widely recognized in politeness research, dating back to Brown and Levinson (1987, p. 172) and their observation that “umms and ahhs and hesitations … are the most salient clue to the presence of an FTA [‘face threatening act’].”
By using hisses and other verbal and nonverbal markers of hesitation, speakers perform a display of reluctance that make their hesitant attitude more conspicuous. Through hissing and performing embodied actions associated with uncertainty, speakers are in effect saying: “Look, I'm being really cautious here, because I know you are not necessarily going to like what I'm going to say next” (cf. Clark and Fox Tree, 2002). Hisses are an effective device for doing this since they are more salient than unfilled pauses (Winter and Grawunder, 2012), or the use of um/uh-type fillers (which occur in Korean speech more than 10 times as frequently as hisses, Kim et al., 2021). Their functions also seem to be more specific to um/uh-type fillers, which occur with disfluencies and, in Korean, as backchannels (Kim et al., 2021). At the same time, using hisses may be preferable over adding more lexical politeness markers, the inclusion of which runs the risks of sounding insincere or trite. By stylizing the sound of their breathing into audible hisses (and accompanying it with shifts in gaze and head alignment), speakers find a way to transform what would otherwise be a silent pause into a noisy and altogether more salient type of pause, which is more noticeable to the interlocutor, and more recognizable as a display of hesitation.
Hisses therefore appear to be more specified in their functions than other types of filler-like sound objects in Korean, or potentially phenomena such as sniffs, clicks and swallowing reported in previous research (Hoey, 2020; Ogden, 2020, 2021). However, as with other sound objects, their interactional meanings are ultimately ambiguous and under-defined. As noted previously by Dingemanse (2020), sound objects derive their interactional utility precisely via their ambiguity and their liminal status between speech and mere noise. Sound objects are “noticeable yet off-record, perceptible yet ignorable” (Dingemanse, 2020, p. 191). Hisses work well as politeness markers precisely because the intentions behind the hiss are open to different interpretations, with each interpretation potentially being deniable. A speaker can hiss to avoid answering a question (such as in examples 4 and 8), without needing to put this reluctance definitely on record. If the speaker is challenged for being reluctant to answer the question, they can always plausibly deny it. Just as Ogden (2020) notes for clicks in English, we can say that hisses in Korean are an effective resource for “audibly not saying something.” They are a way of being reluctant and hesitant, without actually saying on record that you are being reluctant and hesitant.
The increased frequency of hisses and hesitation displays in interactions with status superiors shows us that speakers are trying to show more care and effort when interacting in situations where social distance exists (cf. Winter and Grawunder, 2012). This additional effort that speakers put into appearing reluctant to perform troublesome speech with status superiors makes sense given what we know about Korean language and culture, with social distance encoded directly in grammaticalized honorifics systems in the language (see Sohn, 2001; Brown L., 2015). Moreover, disagreeing with or causing discomfort to superiors is socially rather taboo (Yoon, 2004). This is not necessarily to say that speakers are more reluctant or hesitant when they are addressing superiors, but rather that they are “performing” this hesitation in a more salient way.
This performance of effortful speech is somewhat consistent with previous findings regarding the multimodal way that politeness is expressed toward status superiors in Korean. In terms of body comportment, for instance, Korean speakers have been shown to adopt more erect body positions (Brown and Winter, 2019), which are more effortful than the relaxed postures used with friends. As claimed by Gussenhoven's (2002) “effort code,” performing effortfulness can be used by speakers as a strategy for appearing obliging, which lay lead in turn to perceptions of politeness.
Hisses therefore take on politeness-related functions and can become associated with a more hesitant and careful way of speaking that might be more appropriate for addressing a status superior. It should be noted, however, that any politeness-related meanings communicated by hisses are occurring in a very indirect and context-dependent fashion, particularly given that hisses can also occur in intimate interactions with friends, and/or when the speaker is simply hunting for the correct word (see example 9). They can also be involved in communicating impolite meanings (see example 10 and discussion below). Rather than politeness, it appears that the underlying stance (or first-order indexical meaning, Silverstein, 2003) communicated by hisses is related to difficulty or uncertainty. When used in particular contexts, this leads to the emergence of context-specific meanings (second order indexical meanings) such as hesitancy, reluctance, carefulness, and politeness.
Politeness-related interpretations of hisses may depend in large part on co-occurrence with other politeness-related devices. We saw that hisses for delay and delicacy often appear after the first-person pronominal phrase cey-ka “I-nominative” (examples 3 and 5) which emphasizes a humble stance and which is used when speakers are accepting responsibility (Oh, 2007), as well as other multimodal markers of deference such as bowing and the adoption of a compact body position (example 5). In rapport-threatening speech acts (examples 1 and 2), the speech act itself may be stated with mitigation via the use of non-factive constructions and indirect strategies. Although the content following the hiss may at time feature a decrease in epistemic commitment as claimed by Lee and Lee (2021), we found that it can also contain strong expressions of apology (example 5), for instance, or strong commitments to perform an action, such as offers to redress wrongdoings (example 3).
The fact that politeness-related interpretations of hisses are context-dependent is demonstrated by the occurrence in our data of hisses that convey skepticism. Far from being polite, this is a social action which is open for interpretation as impolite. On one level, these hisses actually have some things in common with the other tokens in our data. They are occurring in disruptive and disjunctive sequences of talk, prior to speakers performing heavily dispreferred and rapport-threatening utterances. However, rather than mitigating the force of the utterance that follows, these hisses seem to emphasize the speaker's skeptical stance toward their interlocutor's utterance and add to the level of impoliteness. Politeness research has shown that even linguistic forms that are heavily biased toward polite meanings such as honorifics (Brown, 2013) or respectful address terms such as Doctor (Bousfield and Johnson, 2007) can be interpreted impolitely when used out of socially normative contexts. In addition, some hisses may not have any specific connection with (im)politeness at all.
It is important to recognize that the findings in this paper were driven by a smaller subset of speakers who are high-frequency hissers, whereas other speakers did not hiss at all. The reasons for this high level of inter-speaker variation are not entirely clear, although it is worth noting that this variation does not exist to the same degree for uh/um-type fillers in Korean (Kim et al., 2021), which suggests that it is something specifically about hisses that is more variable between different speakers. One possible explanation for these inter-individual differences is that despite the ambiguity of hisses, they are still a fairly high-stakes strategy compared with other hesitation options (such as using uh/um-type fillers or unfilled pauses) due to their strong high-frequency component and added saliency. Since any type of filled pauses are known to help listeners make predictions about what is coming next (Clark and Fox Tree, 2002; see Corley and Stewart, 2008), hissing is likely to be quite a strong cue that the speaker is about to say something that is dispreferred. On the one hand, this makes the speaker's polite stance more explicit, but at the same time it also places emphasis on the indiscretion itself. There was also a gender difference in the results with male speakers hissing more frequently than females, although the reasons for this are unclear.
Conclusion
The current paper has shown that in Korean it is indeed polite to hiss, at least in some contexts and for some speakers. Our analysis shows that a subset of our speakers manipulated their frequency of hissing depending on whether they were speaking to a friend or to a status superior, particularly when engaging in sensitive and disruptive segments of speech. Our analysis suggests that hisses mark difficulty and uncertainty, which in context leads to meanings such as hesitancy, reluctance, carefulness, and politeness.
Our results demonstrate that the use of hisses appears clustered at certain politeness hotspots. We take this as evidence that hisses have communicative value during interaction. Results from existent research do indeed suggest that nonverbal speech sounds can trigger communicative expectations on the part of the listener, at least in some circumstances. Of some relevance to the current study, Barr and Seyfeddinipur (2010) found that listeners more strongly expected new referents when the speaker used a recognized filler such as um rather than a random noise. The finding that listeners associate fillers with new or noteworthy information ties in somewhat with our own finding that hisses occur at dispreferred or disjunctive sequences of talk, in other words, when speakers are saying something that is unexpected and/or conversationally disruptive. Ultimately, however, future research focused on listener perception would be needed to fully ascertain the communicative value of hisses in the expression of politeness-related meanings.
Finally, this paper raises the question of whether the use of hisses in Korean as hesitation devices that have politeness functions is something that is specific to the Korean language. Actually, our findings for Korean tie in well with some (rather isolated) observations on the use of other sound objects for politeness-related meanings in previous studies. Harris (2003) observed that British judges and doctors use fillers and hesitation markers to mitigate their speech, whereas Aronsson and Rundström (1989, p. 497) showed that Swedish doctors use hesitation markers such as uhm in pediatric discourse to “soften the doctor's intrusion into the parental authority domain.” The use of hissing noises as hesitation markers is reported only for Danish (Fredsted, 2005). Given the relative lack of research into the detailed social functions of nonverbal speech sounds and the multimodal facets of politeness (see Brown and Prieto, 2017, p. 358), more research is needed to ascertain how different nonverbal speech sounds pattern with the expression of politeness-related meanings across multiple languages.