Abstract
In recent years, the opioid crisis in the United States has sparked significant discussion on doctor-patient interactions concerning chronic pain treatments, but little to no attention has been given to investigating the vocal aspects of patient talk. This exploratory sociolinguistic study intends to fill this knowledge gap by employing prosodic discourse analysis to examine context-specific linguistic features used by the interlocutors of two distinct medical interactions. We found that patients employed both low pitch and creak as linguistic resources when describing chronic pain, narrating symptoms, and requesting opioids. The situational use of both features informs us about the linguistic ways in which patients frame fraught issues like chronic pain in light of the current opioid crisis. This study expands the breadth of phonetic analysis within the domain of discourse analysis, serving to illuminate discussions surrounding the illocutionary role of the lower vocal tract in expressing emotions.
Keywords: Doctor-patient interaction, Opioids, Chronic Pain, Creak, Pitch, Prosody, Discourse Analysis, Sociolinguistics
1. Introduction
Despite growing sociolinguistic interest in the study of intraspeaker voice-quality variation and the role that linguistic choices play in shaping identity and creating social meaning (Mendoza-Denton, 2011; Podesva, 2007; Wilce, 1997), there remains little research in the field of clinical sociolinguistics—the application of linguistic or phonetic techniques to the study of communication in the medical space (Ball et al., 2008). This exploratory study attempts to fill this knowledge gap by conducting prosodic discourse analysis (Chafe, 1993) to describe the different vocal features that patients employ when discussing issues concerning chronic pain and opioids with their physician. We found that patients use both low pitch and creaky voice when (1) narrating symptoms or describing pain and (2) requesting opiates.
Unlike bruises, the symptoms of chronic pain are not always visible, and so patients use “talk”—among other methods such as illustration and demonstration—to describe their suffering and express their complaints and symptoms to the physician (Heath, 2002). Linguistic practices are especially important in such discussions because there are no objective clinical tests to measure chronic pain (Sullivan and Ferrell, 2005). Consequently, clinical assessments and treatment decisions regarding pain are often based solely on the patients’ verbal manifestations of pain and suffering (Burgess et al., 2008; Turk and Okifuji, 1997; Henry and Eggly, 2013).
The current opioid crisis heightens the relevance of investigations focusing on verbal manifestations of pain. In 2011, the United States Center for Disease Control and Prevention (CDC) declared prescription drug abuse as a national epidemic after deaths from accidental overdose exceeded fatalities from vehicular accidents. Because complaints about chronic pain can be interpreted as drug-seeking behavior (Højsted and Sjøgren, 2007), the manner in which patients communicate their chronic pain to physicians may be different from how they discuss other types of pain. Merrill et al. (2002) found fear and mistrust as dominant themes in discussions of opioids for both physicians and patients. Roberts and Kramer (2014) expressed the need for analyzing linguistic practices in medical appointments where pain medications are discussed because issues related to controlled substances are fraught with questions of ethical and moral nature, as well as the potential for abuse and/or the fabrication of symptoms to justify drug-seeking behavior. Matthias et al. (2013) echo this point, calling for more direct explorations to better understand pain management communication. To meet this need, this study offers a phonetic perspective of how patients discuss chronic pain and opioids at a time of crisis. Moreover, because acoustic or prosodic analysis has mostly been used in clinical sociolinguistics to address pathological speech issues (Cernak et al., 2017; Dudy et al., 2018), this research further broadens the applicability of this method by examining actual doctor-patient dialogue.
2. Related Literature
2.1. Behavioral Manifestations of Pain
Pain research seeking to improve doctor-patient communication on chronic pain management is necessary; given that both patients and physicians have reported this dialogue to be challenging and frustrating (Henry et al., 2016). Turk and Okifuji (1997) revealed that neither the severity nor duration of pain solely influence physicians to prescribe opioids. Their findings showed that it is, in fact, the behavioral manifestation of pain—which includes audible expression of distress—that impacts such important healthcare decisions. Subsequent studies on pain management (Hughes et al., 2015; Matthias et al., 2013, 2014) agree that there is value in investigating the prosodic nature of displaying pain in doctor-patient dialogue. This study does just that by bridging voice quality issues to their context-specific discourse functions in medical encounters.
2.2. Voice Quality: Modal versus Creak
Voice quality, as it is used here, refers to the type of phonation speakers produce in the larynx. A speaker, at will, can strategically assume different voice qualities like high or low pitch, whisper, falsetto, breathy, and creak to portray certain attitudes and emotions (Couper-Kuhlen, 2015; Esling, 2012).
Modal phonation refers to the neutral or typical form of speaking, in which speakers talk within the natural pitch range of their voice. The everyday use of the word “pitch” describes the degree of highness or lowness of a tone. Linguistically, pitch is the perception of the rate of vocal fold vibration and that targeting a remarkably low pitch may result in irregularly spaced vocal pulses known as creaky voice or “vocal fry” (Berry, 2001).
Creak and, to a certain degree, low pitch1, are both products of epilaryngeal constriction—a process of the lower vocal tract that results in the shortening, bunching, and adduction of the vocal folds, leading to lower pitch frequencies and, consequently, the shifting from modal to non-modal phonations like creak2 (Esling et al., 2019).
2.3. Interpreting Use of Pitch in Discourse
The introduction of “contextualization cues” in the 1970s—fully interpretable and indexical linguistic signs that possess an embedded context—opened the idea that linguistic styles such as shifting of pitch and use of creak can carry information (Gumperz, 1982; see also Van Dijk 1999, 2011). Later research, including that of Tannen and Wallat (1987: 208), provided evidence that identifiable linguistic cues such as register—”prosodic choices deemed appropriate for the setting and audience”—could be used in the analysis of medical discourse. A speaker’s register can also be described in terms of intonation, the variations in pitch often used in English to express the purpose of utterances as well as understand the attitudes and emotions of other speakers (Couper-Kuhlen, 2015). For example, the use of rising and falling intonation usually helps us understand the difference between statements and questions, among many other ways of regulating discourse.
Earlier studies on pitch and intonation elicited the help of actors to portray emotions because of the idea that neutral speech must also be gathered as a baseline for comparison. Using this method, Zetterholm (1998), as well as Bänziger and Scherer (2005), found that both pitch lowering and creak portray sadness. The design of this study expands on Cruttenden’s (1997) work on intonation, which highlights the idea that a speaker’s register is “marked” when the entire pitch configuration of an utterance is transposed towards the higher or lower limits of their vocal range. In this study, we illustrate “markedness” by using the speaker’s average pitch as a baseline to assess instances in which their pitch deviates significantly from the typical. This method allows for the examination of linguistic style and the information that actual spontaneous speech carries without the need of actors or “neutral speech stimuli.”
2.4. Interpreting Use of Creak in Discourse
Creak is one particular stylistic feature used in creating social meaning that has sparked interest among sociolinguists and discourse analysts. Scholars have studied the relationship between the use of creaky voice and social class (Esling, 1978), gender and sexuality (Henton, 1986; Zimman, 2012, 2013), and stance (Grivicic and Nilep, 2004; Slobe 2018). According to Anderson et al. (2014), the use of creaky voice is becoming increasingly common among young American women and that listeners take the use of creak as untrustworthy and less educated. There are, however, several studies showing that young women’s use of creak is perceived as more dominant and authoritative yet non-aggressive (Borkowska and Pawlowski, 2011; Yuasa, 2010).
While most studies on vocal fry have focused on linguistic attitudes towards young women, there are a few that have examined the contextual application of creak in discourse. More importantly, these studies have looked at the use of voice quality from an intraspeaker perspective —i.e., how an individual employs specific phonation styles in particular discourse contexts—as opposed to the essentializing nature of interspeaker research. Podesva (2007) is among the pioneers in exploring the situational use of creak by taking context, topic, and audience into account. Podesva has demonstrated how a speaker from Long Island named Heath manipulates pitch and strategically employs creak to construct a “diva persona.” Similarly, Mendoza-Denton (2011) analyzed the speech of “Babygirl” and suggested that creak is a discourse-dependent variable employed in the construction of a Chicana hard-of-heart (hardcore) identity. The present study addresses the lack of similar research in realm of medical discourse.
Although creak has been generally understudied in the domain of doctor-patient interactions, there is a key study that provides valuable insight for the present analysis. Notably, Wilce (1997) wrote about the use of creak by Bangladeshi patients to signal weakness, low energy, and misery when interacting with biomedical doctors, herbalists, exorcists, and diviners. Wilce points out that the “markedness” or saliency of creak brings attention to the utterance and lends credibility to the speakers’ reference to their own pain, therefore making vocal fry a learned and internalized social sign that carries particular discourse functions. Similarly, this study pays attention to “marked” speaking turns, in which the speaker’s use of creak and low pitch are perceptually prominent.
3. Methodology
Using prosodic discourse analysis, this study narrows down the different vocal features employed by two patients as they discuss opioids and chronic pain issues with their physicians.
3.1. Participants
Both patients and their respective physicians were recruited from a hospital-based primary care resident clinic in California. Both patients are taking opioid for chronic pain. To account for individual sociolinguistic backgrounds, it is important to note that the participants self-identify as females and are native speakers of American English. Interlocutor information are presented in Table1.
Table 1.
Patient-physician information taken from their pre-consultation questionnaire.
Patient | Resident Physician | ||||
---|---|---|---|---|---|
Pseudonym | Gender | Age Range | Pseudonym | Gender | Age Range |
Patient A | Female | Late 50s | Physician A | Female | Mid 20s |
Patient B | Female | Mid 40s | Physician B | Female | Early 30s |
The primary purpose of patient A’s visit was to request a refill for her opioid prescription after being denied by a different physician due to recent revelations that another pain clinic was already prescribing her with opioids, among other reasons. The physician’s attempt to steer patient A towards alternative means of pain management was met with resistance. On the other hand, Patient B’s reason for scheduling a consultation was to switch to a different type of opioid.
The choice to look at a small number of participants is appropriate in investigations involving language-in-action as it allows for a thorough analysis of the discourse contexts that surround the style shifts in question (Schilling-Estes, 1998). Echoing Podesva’s (2007: 498) perspective on voice quality being a vehicle for social meaning, examining the context-specific use of linguistic variables lends insight into speakers’ intentions and the functionality of their utterances. In this case, analyzing patterns in which patients employ creak and low pitch informs us about their understanding of the discourse functions both variables serve in the medical space.
3.2. Dividing Discourse into Speaking Turns
To adequately and consistently quantify pitch, we separated the data in terms of individual speaking turns, which refers to the entire speech of a specific speaker before another interlocutor mediates and converses. In the following sequence, because of patient A’s quick pause, the physician was able to inject a quick backchannel response. As much as patient A’s second turn is a continuation of the first, we consider them as two separate turns.
P-A: But he wanted me to quit immediately, just like that.
DOC: Yeah.
P-A: And I told him, “I can’t do that. You can’t just quit -- methadone like that.”
It is important to note that “overtalks”—situations in which more than one speaker is talking at once—were excluded from the analysis since they do not provide for an accurate pitch analysis. Brief backchannel responses such as “uh-huh,” “okay,” and “right,” were not considered as well.
3.3. Measuring Pitch and Assessing Creak
Although pitch is the term regularly used in describing the listener’s judgment of what they hear, it is also the perception of the rate of vocal cord vibration in Hertz (cycles per second). In other words, pitch is the subjective attribute of the voice’s fundamental frequency (f0) estimate (Bendor and Wang, 2005). Though certain distinctions exist between the use of both pitch and fundamental frequency (f0), their relationship is established enough to allow us to talk about the speaker’s pitch through f0 measurements (see Gerhard, 2003). This approach enables us to present auditory judgments in a quantitative manner. Pitch or f0 values were obtained using Praat® software (Boersma and Weenik, 2013). Recordings were extracted using Audacity® software (Version 2.1.2, freeware, ©1999–2014. Audacity Team. http://audacity.sourceforge.net/). Background noise was removed through spectral noise gating—a process that works well when the signal in the recording is much louder than the noise3.
Unlike modal utterances, we cannot simply generate accurate pitch values for creak since its f0 is extremely low and the spacing between the pulses are too irregular (Keating et al., 2015). Thus, creaky voice was distinguished from modal voice through perceptual identification followed by examining waveforms and spectrograms to guarantee accuracy. The acoustic analysis in determining creakiness employs the same criteria used by Henton (1986), Gordon and Ladefoged (2001), Podesva (2007), and Mendoza-Denton (2011). Creak is acoustically exemplified by one or more of the following: (1) the irregular spacing of the glottal pulses in wideband spectrographic displays, (2) inconsistencies in f0 values due to slowing of vocal fold vibrations, (3) the abrupt decline in f0, (4) irregularity in the period of each cycle (pitch perturbation or jitter), (5) irregularity in the amplitude of each cycle (amplitude perturbation or shimmer), (6) decreased acoustic intensity relative to modal phonation, and (7) fewer pitch periods per second relative to modal counterpart. For speaking turns that have both modal and creaky segments, only the modal segments were tracked for pitch. This information is summarized in Table 2.
Table 2.
Distribution of speaking turns collected and analyzed.
Speaking Turn | Patient A | Patient B |
---|---|---|
Modal turns (analyzed for pitch) | 166 | 136 |
Entirely creaky (not analyzed for pitch) | 17 | 41 |
Total number of speaking turns | 183 | 177 |
Figure 1 summarizes the acoustic analysis section of our methodology.
Figure 1.
Diagram showing how the data were analyzed acoustically.
The decision to represent pitch through f0 frequency is defensible because this technique can easily be replicated in the medical setting. For instance, if pitch were to be added to patients’ medical notes in order to take records of how they speak in certain contexts, reporting f0 acoustic data would be viable rather than manually noting impressionistic data.
3.4. Coding of Speaking Turns: Pain-Related vs Non-Pain-Related
To point out the linguistic features both patients use to discuss opioids in light of the current crisis, each speaking turn was coded into different contextual categories using a modified version Chronic Pain Coding System (CPCS) developed by Henry et al. (2016). CPCS focuses on the objective characterization of utterances involving pain and opioids, making it appropriate for the current study. There are three main contextual categories included in the two doctor-patient interactions observed: (a) discussions about chronic pain and the opioid medication used to manage it; (b) discussions related to the other types of pain not involving opioids; and (c) discussions that are not about pain. Tables 3, 4, and 5 present the various subcategories under each of the three primary categories mentioned, together with sample excerpts from both patients.
Table 3. (a) Chronic pain (Opioid context).
Discussions about chronic pain and the opioid medication used to manage it.
Patient A | Patient B | |
---|---|---|
Request for opioids | If you could give me methadone pills. I would be happy with that. | I just feel like it [Dilaudid] should be put on there [Pain Contract]. |
Narration/description of chronic pain (Treated with opioids) | I woke up crying last night. I--it’s just still all up here in the shoulder. | I’ve been miserable the last couple days from my neck and shoulders. |
Positive assessment of opioid treatment | I know what is safe for me. I’ve been prescribed with Norco, and it has been working. I’m still alive. | I haven’t been needing Zofran as much since they switched me to that Dilaudid. |
Negative assessment of opioid treatment | None | None |
Ambiguous assessment of opioid | None | (On effectiveness of opioid) It depends on the day, it depends on what I’m doing. |
Opioid-related red flags and threats. | I’ll self-medicate if I have to. I’ll go on the streets. I’m not gonna go through withdrawals. | I probably take more (opioid) overall than I would normally, when I’m in a lot of pain. |
Table 4. (b) Other pain (Non-opioid context).
Discussions about other types of pain that does not involve opioids
Patient A | Patient B | |
---|---|---|
Request for non-opioid medication | I think I need a refill on Dulcolax. | The Robaxin. Um, could you guys change that to 120 tablets a month? |
Unclear requests | So are you gonna prescribe me anything? | None |
Request for information | How many refills do I have on the ibuprofen? | Would it be okay if my daughter picks the prescription up? |
Positive assessment of non-opioid treatment | None | Neurontin does help with some of the um, like my skin or my hair hurting, |
Negative assessment of non-opioid treatment | Ibuprofen is not good for my stomach. | I’ve tried it all. I’ve tried a party bag of ice until I can’t even feel it anymore. |
Ambiguous assessments of non-opioid treatment | None | Robaxin doesn’t really help, but it does |
General agreements | None | Yeah, especially right in here. I would definitely try trigger point injections |
General disagreements | No, I’m not gonna do Tylenol, because it’s not good. | None |
Other pain-related utterances | I’ve had constipation. | I’m kinda worried about asthma. |
Table 5. (c) Non-pain.
Discussions that are not about pain
Patient A | Patient B | |
---|---|---|
Non-pain | We can do Rite-Aid, yeah. | Honestly, it’s just to go to Taco Bell— |
Three coders coded for categories, including a physician and two sociolinguists trained in discourse analysis of medical interactions. The odd number of coders allows for a majority rule in cases of disagreements4. The Fleiss (1971) interrater reliability for these categories by three individual coders is 0.915.
4. Results and Discussions
4.1. Acoustic Analysis
Table 6 shows Patients A and B’s individual average pitch as well as the mean pitch for the three major categories.
Table 6.
Average pitch of all modal speaking turns per category.
Patient A | Patient B | |
---|---|---|
Overall average pitch | 195.298 Hz | 171.108 Hz |
All pain-related utterances | 194.868 Hz | 165.816 Hz |
(a) Chronic pain (Opioid context) | 184.919 Hz | 161.017 Hz |
(b) Other Pain (Non opioid-context) | 197.316 Hz | 166.968 Hz |
(c) Non-pain | 202.791 Hz | 182.552 Hz |
Both patients spoke with lower pitch in utterances that focused on pain, more so if the pain-related discussion involved opioids (including requests and positive assessments of opioids, as well as description of pain in relation to the prescribed opioid). The following diagrams provide a visual representation of each patients’ vocal range and the frequency at which each speaking turn’s average pitch occurs.
As shown in Figure 2, the speakers’ pitch in turns that were about pain and opiates were lower. Patient A’s pain-related utterances that were not in the context of opioids have speaking turns that are distributed towards the upper registers of her speaking voice. Towards the second half of her visit, patient A repeatedly raised her voice and over-enunciated some words for emphasis. Although pitch is not synonymous to volume, patient A’s pitch went up as she intensely argued. Nonetheless, it is in the context of pain and opioids in which she speaks in the lower range of her register. On the other hand, patient B’s use of low pitch is clearer. Table 7 provides a stratified list of results, specifying the average pitch for each subcategories.
Figure 2.
Mean pitch distributions for each of the three coding categories. The width of each violin plot indicates the distribution of the mean pitch located at that point, while the bar shows the overall average for that category.
Table 7.
Average pitch for each subcategory
Average Pitch (in Hz) | ||
---|---|---|
Patient A | Patient B | |
Baseline: overall average pitch | 195.298 | 171.108 |
(a) Chronic pain (Opioid context) | ||
Request for opioids | 175.372* | 150.993* |
Narration/description of chronic pain | 173.256* | 152.509* |
Positive assessment of opioids | 200.242 | 175.799 |
Negative assessment of opioids | -- | -- |
Ambiguous assessment of opioids | -- | 175.629 |
Opioid-related red flags and threats. | 202.320 | 151.908* |
(b) Other pain (Non-opioid context) | ||
Request for non-opioid medication | -- | 167.327* |
Unclear requests | 208.608 | -- |
Request for information | 195.632 | 184.222 |
Positive assessment of non-opioid treatment | -- | 154.733* |
Negative assessment of non-opioid treatment | 198.157 | 180.428 |
Ambiguous assessment of non-opioid treatment | -- | 180.394 |
General agreements | -- | 164.302* |
General disagreements | 201.411 | -- |
Other pain-related utterances | 196.490 | 164.747* |
(c) Non-pain | ||
Non-pain | 202.791 | 182.552 |
Asterisk indicates that the average pitch of that category is lower than the speaker’s overall average pitch.
After narrowing down the categories, we found that both patients had the lowest pitch when narrating or describing chronic pain or asking for the opioid medication they use to remedy that pain. Figure 3 demonstrates mean pitch distributions for both patients in the following discourse contexts: (i) request for opioid treatment and (ii) description/narration of pain.
Figure 3.
Mean pitch distributions of opioid requests and pain narration separated from the rest of the speaking turns. The width of each violin plot represents the pitch distribution located at that point while the bar represents the average pitch for that category.
It is important to note that the lowering of both patients’ pitch is not restricted to only the concluding segments of a speaking turn. Low pitch also occurs in other parts of the turn and are mostly sustained once initiated. Moreover, there is no correlation between the length of a turn and the potential for pitch lowering.
Examples of instances in which both patients lower their pitch when discussing chronic pain and opioids are shown in Figures 4a and 4b below.
Figure 4a. Example of pain narration from Patient A.
Waveform and spectrogram of a turn in which patient A discusses pain. The dense line on the spectrogram represents pitch. The mean pitch of the present turn is below her overall average and covers only the lower 22 percent of her vocal range.
Figure 4b. Example of pain narration from Patient B.
Waveform and spectrogram of a turn in which patient B discusses pain. The dense line on the spectrogram represents pitch. The mean pitch of the present turn is below her overall average and covers only the lower 20 percent of her vocal range.
In what follows, we present our findings on the use of non-modal phonation, creak.
4.2. Entirely Creaky Speaking Turns
Although pitch cannot be measured for creak, we still coded the entirely creaky speaking turns to study the context in which the feature is used by both patients (See Table 8).
Table 8.
Distribution of completely creaky speaking turns
Number of creaky turns | ||
---|---|---|
Patient A | Patient B | |
Context: Pain | 11 | 35 |
Context: Non-pain | 6 | 6 |
None of the completely creaky turns were measured for pitch.
Of all of the creaky turns, 65 percent of Patient A’s and 85 percent of Patient B’s were about pain. Figures 5 and 6 provide examples of waveforms and spectrograms of words articulated in both creaky and modal phonation. At the bottom of each figure are waveforms zooming into the beginning of the vowel to show the differences in periodicity.
Figures 5 and 6. Creaky and modal versions of the same word.
Waveforms and spectrograms illustrating the creak and modal versions of the word “Pain” and “Bad” spoken by Patients A and B, respectively. The line on the spectrogram refers to pitch. The waveform for the beginning of the vowel is also presented.
The creak and modal examples above adhere to the characteristics outlined in the methodology section. Both creaky examples have aperiodic glottalization, lower acoustic intensity, and significantly lower amplitudes in comparison to their modal counterparts. The spacing between pulses is more regular with the modal examples. Figure 6a also has alternating longer and shorter pulses that is often characteristic of creaking (Keating et al., 2015). Lastly, the pitch values are irregular in Figure 5a and completely undetectable in Figure 6a.
5. Discussion
Our study demonstrates that pitch or f0 values were lower and the use of creak was more apparent when both patients discussed chronic pain and opiates. Because both phonation styles are associated with the lower vocal tract and the use of lower register, it is not unexpected that creaky segments are detected in utterances with low modal pitch, as shown in the examples below.
In excerpt (1), patient A informs the doctor about not being prescribed with opioids by another physician. The primary reasons for the denial include patient A’s toxicology results and the fact that she is already taking another opioid from another pain clinic. (The numbers in parenthesis represent the mean pitch of each turn’s modal segment while utterances in bold denote creak).
(1)
Physician: Alright. So -- um -- alright, let’s start with what you would like to talk about.
Patient A (139.9 Hz): Okay, and I was very upset with my last visit with Dr. <name>.
Physician: M-mm.
Patient A (136.1 Hz): He wouldn’t prescribe any an- -- any meds, any pain medication that day, so I’ve been without Norco for almost two months now.
Physician: Hm-mm.
Patient A (181.4 Hz): I’ve been in extreme pain, with my shoulder that is still hurting. I can’t do physical therapy, because my -- I might even need to have surgery. I wanna see the s-, the surgeon again, because it’s just not healing.
Physician: Hm-mm.
Patient A (169.5 Hz): It’s not healing at all. And, I mean, he’s with me day and night practically when he’s not at work, and for the past four years we’ve been together, and he can verify -- um -- the pain that I’m going through -- with all this.
Lines 6 and 8 are examples of turns in which patient A describes her pain using low pitch and creak. In both turns, the patient’s pitch is seven and thirteen percent lower, respectively, compared to her overall average of 195.298 Hz. In line 4, patient A proves that not all requests have to take the interrogative form. According to Robinson (2001), implicit or indirect requests can take any grammatical form as long as the utterance performs its soliciting function in the context of the medical visit. Line 4 is coded as an opiate request because the purpose of the turn was not only to inform the physician that no opiate has been prescribed but also to suggest that the situation must be addressed. In this turn, the patient’s pitch is 30 percent lower than the baseline.
The pairing of low pitch and creak in opiate requests is also discernible in patient B’s speech as shown in excerpt (2) below. In the beginning of this exchange, patient B asks to be prescribed with Cymbalta, Neurontin, and Lyrica, none of which are opiates. She proceeds to talk about the main reason for her visit, which is to get the opiate, Dilaudid, added into her pain contract.
(2) (Utterances in bold denote creaky voice)
Physician: How can I help you today?
Patient B (166.0 Hz): Uh, there’s a few things. Uh, the Cymbalta that—I don’t remember his name, but the last doctor I saw—put me on the Cymbalta.
Physician: Yeah.
Patient B (152.6 Hz): Uh, I don’t have any more and there was [sic] no refills, so -- I’ve still got that.
Physician: I can certainly refill that for you if that seems to be helping.
Patient B (160.7 Hz): Okay. Yeah, I think it does help a little bit with the anxiety and stuff. I don’t know if it’s helping with the pain, um—
Physician: The, the effect on the pain might be sort of subtle... it may, it may be helping to reduce the amount of other pain medicines you require.
Patient B (172.2 Hz): Unless—my—cuz [sic] I did pretty good, I think—- with the Dilaudid they did give me. And also, uh, I took the Neurontin three times a day. I really have a hard time remembering to do that three times a day all the time.
Physician: Uh huh.
Patient B (184.5 Hz): And I, I have taken Lyrica before. It was just a two-week trial...
In the example above, patient B lists four medications she wants to be filled, yet it’s worth noting that creak was only employed in the discussion of pain and Dilaudid in lines 6 and 8. In line 6, patient B introduces the idea that Cymbalta does not address her chronic pain, which consequently leads to the reveal that it is the opiate Dilaudid that helps in line 8. The comments on lines 6 and 8 are coded as requests—implicitly delivered through negative evaluation of a non-opiate drug in line 6 and a positive assessment of the opiate in line 8. Starting in line 10 up until opiates are discussed again, patient B goes back to speaking with her regular modal voice, making the use of low pitch on discussions about controlled substances and chronic pain more apparent.
Irvine (2001) suggests that we could interpret the motivations behind the use of certain linguistic styles by examining situations in which it is absent. We found that requests for non-opiate medication have no semblance of low pitch and creak, as shown in the example below:
Patient B (193.4 Hz): Oh, the Robaxin. Um, could you guys change that from 100 to 120 tablets for a month?
In this excerpt, the request for the muscle relaxant, Robaxin, takes the conventional interrogative form. Such straightforward request contrasts the suggestive opiate requests presented earlier. In fact, both patients started the discussion on opioids by “reporting” its effectiveness while simultaneously highlighting the presence and severity of their chronic pain. According to Robinson (2001) and Gill et al. (2001), reporting—in the form of assessments—is used by patients to implicitly make sensitive requests like asking for addictive medications, without revealing their position towards the request5.
Excerpt (3) below illustrates an exchange in which patient B neither uses creak nor lowers her pitch to discuss a condition that is not addressed by an opioid.
(3)
Physician: Anything else that uh I can help you with today?
Patient B (193.0 Hz): Uh, my asthma. I live in Sunnyville, and it was pretty much fogged in smoke.
Physician: Okay.
Patient B (190.5 Hz): And it started bothering my asthma right away.
Physician: Oh.
Patient B (194.7 Hz): My sat was 95, but—I’ve been using my inhaler a lot.
Our data show that both patients find value in switching to a distinct register when discussing chronic pain and opiates. Going back to Cruttenden’s (1997) notion about register, the salient shifting in pitch indicates emphasis to what is being said. Considering that both patients overwhelmingly use low pitch and creak in very specific contexts tells us that both linguistic variables are being employed for stylistic work. Specifically, both phonation styles serve as pragmatic resources used to express pain as well as to request the medication that they believe best manages their misery.
5.1. The Nature of Opioid-related Utterances
These findings raise a significant question: what is distinct about the topics of chronic pain and opiates that motivate the change in vocal style? The opioid crisis has brought stigma to the discussion of controlled substances, which has made requesting opioids a fraught process within the medical setting. We already witness such effect in this study, through both patients’ suggestive framing of requests. Roberts and Kramer (2014) found that patients orient issues surrounding pain medications as problematic, morally suspect, and easily refusable. Patients have to confront the effects of the opioid crisis by increasing their sensitivity towards the potential concerns that doctors may have about dependency and addiction when prescribing opioids. Frequently, patients find it necessary to defend their moral character and present themselves as credible, responsible, and aware when the topic and requests are sensitive, challenging, have high chances of denial, potentially controversial, and morally fraught. It is evident from our findings that such tasks are accomplished by both patients through the situational use of low pitch and creak.
What sets opioid-related conversations further apart is the fact that it could be a source of disagreement because patients and physicians often do not share the same priorities when it comes to managing chronic pain (Henry et al., 2017). In fact, a post-visit survey given to the participants in this study reveal that both patients ranked “reducing pain intensity” as their most important goal while their respective physicians placed higher emphasis on improving the patient’s overall function. Alerting physicians about the use of specific vocal features in disagreeable discussions like opioids, could alert them to confront the disagreement by returning the conversation to goal setting.
Clearly, our findings show that the dynamics involved in medical encounters involving chronic pain and opiates are indeed different from other primary care visits. If the entirety of patient A’s appointment were about her constipation while patient B’s were about her asthma, the linguistic practices they would employ throughout their consultation would less likely involve creak and pitch lowering, as suggested by the way they discussed these same concerns in this study.
5.2. Discourse Functions: Expressiveness of Low Pitch and Creak
Patients cannot always show visible evidence of chronic pain, therefore their only recourse is to express their symptoms using the primary activity that takes place in medical interactions: talking. This raises the question as to why creak and low pitch are used in conjunction with requests for opiates. What do these variables index? We already know that low pitch/f0 and creak collaboratively portrays sadness, misery, weakness, certainty, and credibility (Bänziger and Scherer, 2005; Borkowska and Pawlowski, 2011; Wilce, 1997; Yuasa, 2010). According to Podesva (2007), the common denominator that describes the polysemic functions carried by a linguistic variable is its “expressiveness”—indexed within the particular discourse contexts in which the variable is repeatedly employed. As such, our analysis of the conversational contexts suggests that the discourse functions low pitch and creak serve include: (a) addressing medical issues that are delicate, important, morally fraught, bound to receive an assessment, and could possibly be refuted; (b) self-reporting chronic pain symptoms, which could be difficult to prove and easily questioned due to lack of methods to properly evaluate it; and (c) requesting addictive painkillers that could easily be refused or interpreted as drug-seeking behavior.
6. Conclusion
This exploratory study examined important linguistic features present in two doctor-patient interactions on chronic pain management. Through discourse analysis, we found that the patients lowered their pitch and employed creak when discussing chronic pain and opioids with their physicians. Both features are associated outcomes of epilaryngeal constriction, which in itself carries paralinguistic functions (Moisik, 2013). This study expands the breadth of phonetic analysis within the domain of discourse analysis, leading to the need to explore topics involving the illocutionary role of the lower vocal tract in expressing emotions.
Studying micro-linguistic practices in the medical space may improve the overall efficacy of health communication, which is vital for setting shared pain management goals and reducing inappropriate opioid prescribing (Henry et al., 2017). The situational use of low pitch and creak tells us about how patients frame conversations on opioids in light of the current medical climate. Patients, regardless of intention, are tasked with navigating the challenging and fraught discussion of opioids, knowing well that their requests could be refused or their symptoms questioned. Awareness of the discourse functions of low pitch and creak during discourse may alert physicians about the possible concerns of patients. Recognizing such discernible shift in register could also signal the physician into bringing the discussion back towards their shared goals.
Additional investigations would allow us to discover whether men use the same vocal features when interacting with doctors. Lastly, examining the correlation between auditory judgments and prescribing decisions through a perception test could also further this study.
Acknowledgements:
We would like to express our sincere gratitude to Teun Van Dijk and anonymous reviewers for their expert feedback. Their incisive and thoughtful comments helped reconceptualize several parts of this exploration. Of course, all remaining faults are our own.
Funding: The videos analyzed in this study were collected with funding from the National Institutes of Health (grants KL2TR000134 and UL1TR000002) and the University of California, Davis Department of Internal Medicine.
Appendix A: Coding Scheme
R1A | Request for Non-Opioid treatment |
R1B |
Request for Opioid treatment Request for opioid increase. Request for routine opioid refill. Patient request for opioid switch. |
R1C | Unclear request regarding Prescription |
R2 |
Request for information and logistics Request related to logistics. Patient requests for information. Patient request for non-opioid pain treatments. |
T2D |
Opioid related threats/red flags Patient statements about serious opioid side effects and red flags. Statements about less serious opioid side effects. Opioid-related threats. |
E2 | Narration of pain and description of symptoms |
T1A | Positive assessments /satisfaction with non-opioid treatments. |
T1B | Negative assessments / dissatisfaction with non-opioid treatments. |
T1C | Expressions of uncertainty or ambiguity about a non-opioid treatment plan. |
T2A | Positive assessments / satisfaction with opioids. |
T2B | Negative assessments /dissatisfaction with opioids. |
T2C | Expressions of uncertainty or ambiguity about opioids. |
N1 | Disagreement or resistance. |
N3 | Agreement |
X | Non Pain Related Utterances |
H | Other Pain Related Utterances |
Appendix B: Pitch Tracking Data for Patients A and B (Summaries at the end)
Patient A
Speaking Turn | Context | Mean | Min | Max | Range | Opioid Context | Pain Context |
---|---|---|---|---|---|---|---|
9 | H | 164.783 | 120.73 | 208.855 | 88.125 | ✔ | |
12 | H | 163.907 | 130.007 | 204.602 | 74.595 | ✔ | |
H | 184.007 | 128.17 | 246.504 | 118.334 | ✔ | ||
14 | H | 184.689 | 135.189 | 228.14 | 92.951 | ✔ | |
H | 186.901 | 138.096 | 214.035 | 75.939 | ✔ | ||
H | 183.109 | 125.165 | 300.135 | 174.97 | ✔ | ||
16 | H | 179.63 | 145.518 | 218.222 | 72.704 | ✔ | |
22 | R1B | 136.113 | 119.113 | 283.215 | 164.102 | ✔ | ✔ |
24 | E2 | 177.309 | 125.172 | 246.814 | 121.642 | ✔ | ✔ |
24 | E2 | 185.491 | 135.693 | 233.374 | 97.681 | ✔ | ✔ |
26 | E2 | 169.521 | 136.813 | 204.808 | 67.995 | ✔ | ✔ |
28 | H | 174.096 | 129.277 | 219.569 | 90.292 | ✔ | |
30 | H | 173.692 | 121.563 | 264.423 | 142.86 | ✔ | |
32 | H | 193.155 | 157.114 | 262.798 | 105.684 | ✔ | |
34 | H | 183.187 | 158.745 | 214.013 | 55.268 | ✔ | |
45 | H | 197.092 | 115.095 | 324.744 | 209.649 | ✔ | |
47 | H | 200.927 | 149.362 | 335.386 | 186.024 | ✔ | |
49 | H | 191.869 | 115.343 | 298.943 | 183.6 | ✔ | |
53 | H | 192.296 | 150.457 | 227.269 | 76.812 | ✔ | |
67 | H | 206.175 | 134.382 | 303.281 | 168.899 | ✔ | |
69 | H | 186.492 | 147.243 | 255.286 | 108.043 | ✔ | |
71 | H | 195.17 | 144.291 | 274.186 | 129.895 | ✔ | |
73 | H | 203.499 | 169.799 | 249.01 | 79.211 | ✔ | |
75 | H | 190.273 | 144.114 | 253.458 | 109.344 | ✔ | |
102 | H | 190.403 | 144.018 | 251.154 | 107.136 | ✔ | |
104 | H | 173.239 | 125.369 | 246.372 | 121.003 | ✔ | |
106 | H | 214.232 | 142.61 | 266.435 | 123.825 | ✔ | |
108 | R1B | 182.516 | 139.534 | 243.977 | 104.443 | ✔ | ✔ |
110 | R1B | 157.005 | 137.904 | 198.482 | 60.578 | ✔ | ✔ |
118 | R1B | 184.221 | 159.142 | 280.263 | 121.121 | ✔ | ✔ |
120 | R1B | 185.716 | 155.096 | 251.27 | 96.174 | ✔ | ✔ |
122 | R1B | 187.532 | 155.309 | 306.435 | 151.126 | ✔ | ✔ |
124 | H | 205.96 | 134.761 | 301.223 | 166.462 | ✔ | |
126 | H | 173.122 | 152.454 | 273.56 | 121.106 | ✔ | |
128 | H | 186.943 | 154.141 | 239.359 | 85.218 | ✔ | |
130 | H | 210.177 | 142.305 | 265.149 | 122.844 | ✔ | |
132 | E2 | 168.11 | 111.353 | 246.173 | 134.82 | ✔ | ✔ |
136 | H | 210.058 | 163.922 | 248.125 | 84.203 | ✔ | |
138 | E2 | 149.553 | 130.15 | 165.406 | 35.256 | ✔ | ✔ |
140 | H | 193.969 | 143.913 | 256.311 | 112.398 | ✔ | |
142 | H | 189.595 | 130.251 | 212.011 | 81.76 | ✔ | |
144 | H | 192.092 | 167.696 | 228.539 | 60.843 | ✔ | |
146 | H | 171.283 | 121.216 | 222.346 | 101.13 | ✔ | |
148 | H | 192.422 | 146.019 | 259.952 | 113.933 | ✔ | |
158 | H | 183.827 | 158.028 | 216.97 | 58.942 | ✔ | |
169 | H | 196.024 | 157.312 | 233.51 | 76.198 | ✔ | |
171 | H | ✔ | |||||
173 | H | ✔ | |||||
186 | X | ||||||
188 | X | 206.509 | 125.535 | 221.292 | 95.757 | ||
193 | X | ||||||
208 | X | ||||||
216 | X | 209.451 | 142.896 | 256.617 | 113.721 | ||
218 | H | 189.623 | 160.729 | 212.696 | 51.967 | ✔ | |
221 | X | ||||||
231 | X | ||||||
239 | H | 203.089 | 122.546 | 317.979 | 195.433 | ✔ | |
247a | R2 | 205.438 | 140.842 | 259.43 | 118.588 | ✔ | |
247b | R2 | ✔ | |||||
254 | H | 193.994 | 167.462 | 214.392 | 46.93 | ✔ | |
260 | H | 203.373 | 155.662 | 313.438 | 157.776 | ✔ | |
262 | H | 197.593 | 168.619 | 232.01 | 63.391 | ✔ | |
271 | H | 186.351 | 115.262 | 252.639 | 137.377 | ✔ | |
273 | X | 201.51 | 143.765 | 244.479 | 100.714 | ||
275 | X | 203.438 | 136.561 | 216.375 | 79.814 | ||
279 | X | ||||||
303 | H | ✔ | |||||
305 | H | ✔ | |||||
307 | H | ✔ | |||||
311 | H | ✔ | |||||
313 | H | ✔ | |||||
315 | H | ✔ | |||||
317 | H | ✔ | |||||
319 | H | ✔ | |||||
325 | H | 168.923 | 115.235 | 220.667 | 105.432 | ✔ | |
341 | H | 179.78 | 129.131 | 219.357 | 90.226 | ✔ | |
357 | R1C | 199.216 | 154.459 | 270.33 | 115.871 | ✔ | |
359 | T2D | 190.259 | 145.832 | 238.316 | 92.484 | ✔ | ✔ |
361 | T1B | 190.415 | 157.972 | 242.936 | 84.964 | ✔ | |
363 | H | 192.312 | 157.468 | 273.975 | 116.507 | ✔ | |
365 | H | 195.694 | 147.233 | 274.21 | 126.977 | ✔ | |
369 | H | 204.48 | 126.311 | 285.751 | 159.44 | ✔ | |
371 | H | 215.318 | 161.889 | 270.477 | 108.588 | ✔ | |
373 | H | 203.487 | 168.835 | 253.628 | 84.793 | ✔ | |
375 | H | 219.21 | 141.131 | 279.262 | 138.131 | ✔ | |
377 | T2D | 200.914 | 143.658 | 250.758 | 107.1 | ✔ | ✔ |
379 | H | 219.21 | 131.131 | 289.22 | 158.089 | ✔ | |
381 | R1B | 180.93 | 146.658 | 228.98 | 82.322 | ✔ | ✔ |
383 | E2 | 167.792 | 127.175 | 232.103 | 104.928 | ✔ | ✔ |
387 | E2 | 169.769 | 132.357 | 207.39 | 75.033 | ✔ | ✔ |
389 | H | 197.88 | 145.9 | 289.67 | 143.77 | ✔ | |
399 | T2A | 187.040 | 142.160 | 235.904 | 93.744 | ✔ | ✔ |
401 | H | 181.878 | 155.250 | 252.788 | 97.538 | ✔ | |
407 | H | 200.035 | 124.344 | 253.802 | 129.458 | ✔ | |
411 | H | 199.921 | 161.501 | 267.553 | 106.052 | ✔ | |
425 | H | 200.073 | 130.863 | 284.587 | 153.724 | ✔ | |
427 | H | 199.385 | 156.325 | 290.202 | 133.877 | ✔ | |
443 | X | 200.899 | 140.533 | 238.628 | 98.095 | ||
447 | X | 202.271 | 143.439 | 247.937 | 104.498 | ||
451 | X | 198.665 | 137.752 | 287.606 | 149.854 | ||
453 | X | 199.63 | 168.179 | 291.13 | 122.951 | ||
491 | H | 180.386 | 159.74 | 263.798 | 104.058 | ✔ | |
493 | H | 169.453 | 134.69 | 231.948 | 97.258 | ✔ | |
495 | N1 | 201.775 | 155.35 | 279.71 | 124.36 | ✔ | |
499 | H | 192.941 | 145.153 | 295.153 | 150 | ✔ | |
503 | N1 | 196.534 | 149.131 | 253.357 | 104.226 | ✔ | |
505 | T1B | 181.846 | 155.153 | 257.07 | 101.917 | ✔ | |
507 | H | 180.464 | 165.953 | 253.843 | 87.89 | ✔ | |
509 | H | 202.224 | 156.481 | 274.004 | 117.523 | ✔ | |
511 | T2D | 187.441 | 143.887 | 251.711 | 107.824 | ✔ | ✔ |
513 | T2D | 192.885 | 136.668 | 255.098 | 118.43 | ✔ | ✔ |
517 | H | 203.678 | 134.614 | 287.051 | 152.437 | ✔ | |
521 | H | 197.946 | 163.413 | 286.381 | 122.968 | ✔ | |
523 | H | 228.889 | 154.041 | 373.66 | 219.619 | ✔ | |
525 | H | 207.04 | 162.16 | 305.904 | 143.744 | ✔ | |
527 | T2D | 199.972 | 155.253 | 292.426 | 137.173 | ✔ | ✔ |
529 | T1B | 208.439 | 155.087 | 296.758 | 141.671 | ✔ | |
531 | N1 | 203.858 | 156.22 | 278.364 | 122.144 | ✔ | |
533 | E2 | 180.552 | 128.571 | 218.881 | 90.31 | ✔ | ✔ |
535 | H | 210.425 | 151.954 | 302.841 | 150.887 | ✔ | |
537 | H | 219.52 | 138.365 | 276.878 | 138.513 | ✔ | |
539 | R1C | 212.914 | 156.313 | 302.98 | 146.667 | ✔ | |
541 | R1C | 196.888 | 134.199 | 302.186 | 167.987 | ✔ | |
543 | N1 | 195.955 | 134.43 | 288.258 | 153.828 | ✔ | |
545 | H | 211.177 | 174.251 | 294.834 | 120.583 | ✔ | |
547 | T2D | 204.082 | 174.684 | 285.723 | 111.039 | ✔ | ✔ |
549 | H | 198.736 | 173.114 | 290.267 | 117.153 | ✔ | |
551 | R1B | 188.943 | 130.45 | 265.995 | 135.545 | ✔ | ✔ |
553 | N1 | 201.676 | 166.885 | 292.605 | 125.72 | ✔ | |
555 | H | 213.585 | 140.02 | 294.485 | 154.465 | ✔ | |
557 | H | 184.469 | 132.948 | 225.704 | 92.756 | ✔ | |
559 | R2 | 187.834 | 152.264 | 221.348 | 69.084 | ✔ | |
562 | R1C | 225.412 | 177.142 | 285.305 | 108.163 | ✔ | |
564 | N1 | 203.757 | 150.412 | 292.27 | 141.858 | ✔ | |
566 | H | 208.579 | 138.25 | 288.655 | 150.405 | ✔ | |
568 | T1B | 203.338 | 135.995 | 300.237 | 164.242 | ✔ | |
570 | R2 | 213.39 | 154.772 | 284.089 | 129.317 | ✔ | |
572 | H | 235.596 | 156.295 | 310.296 | 154.001 | ✔ | |
574 | H | 214.883 | 169.737 | 313.636 | 143.899 | ✔ | |
576 | H | 221.902 | 152.495 | 279.186 | 126.691 | ✔ | |
578 | H | 220.564 | 169.27 | 292.588 | 123.318 | ✔ | |
596 | E2 | 179.961 | 119.002 | 215.887 | 96.885 | ✔ | ✔ |
598A | E2 | 170.218 | 138.593 | 204.603 | 66.01 | ✔ | ✔ |
598B | T2D | 244.784 | 158.553 | 397.457 | 238.904 | ✔ | ✔ |
600 | H | 202.84 | 171.158 | 298.37 | 127.212 | ✔ | |
602 | H | 214.079 | 153.633 | 290.977 | 137.344 | ✔ | |
604 | T2A | 213.443 | 148.843 | 282.315 | 133.472 | ✔ | ✔ |
606 | T2D | 200.694 | 181.216 | 282.63 | 101.414 | ✔ | ✔ |
608 | H | 190.604 | 155.014 | 287.312 | 132.298 | ✔ | |
610 | H | 191.14 | 163.84 | 261.937 | 98.097 | ✔ | |
614 | N1 | 180.394 | 152.649 | 267.89 | 115.241 | ✔ | |
618 | N1 | 198.061 | 177.285 | 302.704 | 125.419 | ✔ | |
620 | N1 | 214.741 | 174.543 | 292.981 | 118.438 | ✔ | |
622 | H | 199.76 | 171.271 | 302.349 | 131.078 | ✔ | |
626 | T1B | 206.748 | 162.679 | 290.65 | 127.971 | ✔ | |
628 | E2 | 187.54 | 142.816 | 260.246 | 117.43 | ✔ | ✔ |
630 | N1 | 199.482 | 158.285 | 289.327 | 131.042 | ✔ | |
632 | H | 206.054 | 178.051 | 308.142 | 130.091 | ✔ | |
634 | H | 204.211 | 176.155 | 311.083 | 134.928 | ✔ | |
636 | H | 209.86 | 158.6 | 311.024 | 152.424 | ✔ | |
638 | H | 207.745 | 166.702 | 314.258 | 147.556 | ✔ | |
642 | H | 192.382 | 154.829 | 292.289 | 137.46 | ✔ | |
644 | H | 183.437 | 173.388 | 281.969 | 108.581 | ✔ | |
646 | R2 | 199.755 | 163.888 | 267.17 | 103.282 | ✔ | |
654 | H | 212.745 | 166.702 | 314.258 | 147.556 | ✔ | |
658 | X | 202.745 | 146.702 | 242.258 | 95.556 | ||
660 | R2 | 202.366 | 145.511 | 304.686 | 159.175 | ✔ | |
664 | H | 182.366 | 145.511 | 244.686 | 99.175 | ✔ | |
674 | N1 | 219.293 | 157.325 | 282.693 | 125.368 | ✔ | |
676 | H | 205.766 | 174.508 | 310.662 | 136.154 | ✔ | |
678 | H | 203.675 | 143.088 | 316.62 | 173.532 | ✔ | |
680 | T2D | 218.837 | 151.444 | 306.27 | 154.826 | ✔ | ✔ |
686 | H | 207.516 | 137.65 | 271.281 | 133.631 | ✔ | |
688 | H | 219.621 | 197.805 | 279.329 | 81.524 | ✔ | |
690 | H | 189.258 | 168.801 | 297.935 | 129.134 | ✔ | |
692 | H | 180.787 | 160.833 | 198.839 | 38.006 | ✔ | |
696 | H | 190.905 | 180.207 | 273.403 | 93.196 | ✔ | |
698 | H | 198.72 | 156.461 | 290.714 | 134.253 | ✔ | |
700 | H | 190.164 | 169.417 | 282.201 | 112.784 | ✔ | |
702 | R2 | 187.912 | 149.705 | 226.951 | 77.246 | ✔ | |
704 | R2 | 184.456 | 133.748 | 229.723 | 95.975 | ✔ | |
706 | R2 | 183.906 | 143.671 | 205.45 | 61.779 | ✔ | |
708 | T2D | 183.333 | 170.926 | 283.102 | 112.176 | ✔ | ✔ |
Shaded rows denote entirely creaky speaking turn.*
Patient B
Speaking Turn |
Context | Mean | Min | Max | Range | Opioid Context |
Pain Context |
---|---|---|---|---|---|---|---|
8 | X | 166.005 | 140.428 | 235.628 | 95.2 | ||
10 | H | 152.624 | 125.386 | 173.289 | 47.903 | ✔ | |
12 | R1A | 138.099 | 121.188 | 159.123 | 37.935 | ✔ | |
16 | H | 199.853 | 100.725 | 305.317 | 204.592 | ✔ | |
18 | T1C | 160.675 | 134.325 | 186.21 | 51.885 | ✔ | |
26 | T2A | 177.498 | 157.108 | 198.591 | 41.483 | ✔ | ✔ |
28 | T2A | 172.222 | 141.605 | 300.588 | 158.983 | ✔ | ✔ |
30 | H | 179.623 | 138.108 | 241.546 | 103.438 | ✔ | |
32.1 | H | 184.535 | 100.304 | 287.728 | 187.424 | ✔ | |
32.2 | H | ✔ | |||||
34 | H | 145.018 | 114.272 | 162.64 | 48.368 | ✔ | |
36 | R2 | 176.542 | 135.526 | 197.539 | 62.013 | ✔ | |
38 | R1B | 162.562 | 122.836 | 194.049 | 71.213 | ✔ | ✔ |
40 | H | ✔ | |||||
42 | H | 158.633 | 119.406 | 230.721 | 111.315 | ✔ | |
44 | H | ✔ | |||||
54 | H | 141.135 | 118.557 | 156.776 | 38.219 | ✔ | |
56 | H | ✔ | |||||
58 | H | 177.983 | 122.46 | 195.193 | 72.733 | ✔ | |
60.1 | H | 161.995 | 101.918 | 261.306 | 159.388 | ✔ | |
60.2 | H | 129.865 | 106.853 | 152.301 | 45.448 | ✔ | |
64 | T1C | 178.737 | 119.766 | 285.424 | 165.658 | ✔ | |
66 | H | 139.418 | 123.954 | 154.07 | 30.116 | ✔ | |
76 | H | 156.025 | 111.376 | 187.743 | 76.367 | ✔ | |
78 | H | 154.524 | 115.797 | 368.434 | 252.637 | ✔ | |
80 | H | 147.852 | 134.949 | 177.544 | 42.595 | ✔ | |
82.1 | T1A | 179.491 | 104.647 | 288.368 | 183.721 | ✔ | |
82.2 | T1A | 151.381 | 109.638 | 261.261 | 151.623 | ✔ | |
88 | T1B | 155.842 | 121.583 | 181.912 | 60.329 | ✔ | |
90 | X | 167.448 | 103.391 | 170.093 | 66.702 | ||
94 | X | 176.489 | 108.333 | 183.876 | 75.543 | ||
98 | X | 168.945 | 132.691 | 185.592 | 52.901 | ||
102 | X | 214.178 | 147.974 | 308.762 | 160.788 | ||
104 | X | 194.182 | 101.334 | 326.326 | 224.992 | ||
106 | X | 153.429 | 100.17 | 199.934 | 99.764 | ||
108 | X | 162.764 | 114.468 | 163.471 | 49.003 | ||
110 | X | 160.113 | 114.763 | 170.869 | 56.106 | ||
122 | N3 | ✔ | |||||
126 | T1A | ✔ | |||||
128 | R1B | 146.012 | 121.302 | 195.268 | 73.966 | ✔ | ✔ |
130 | R1B | ✔ | ✔ | ||||
132 | T2A | 168.296 | 127.908 | 208.535 | 80.627 | ✔ | ✔ |
142 | H | 214.225 | 117.833 | 302.108 | 184.275 | ✔ | |
144 | R1B | 165.902 | 106.659 | 298.712 | 192.053 | ✔ | ✔ |
148 | H | 166.445 | 111.287 | 243.513 | 132.226 | ✔ | |
150 | R1B | ✔ | ✔ | ||||
156 | T2A | 185.18 | 143.841 | 278.464 | 134.623 | ✔ | ✔ |
158 | H | ✔ | |||||
160 | H | ✔ | |||||
162 | H | 182.424 | 140.498 | 210.107 | 69.609 | ✔ | |
168 | H | 195.336 | 133.528 | 277.355 | 143.827 | ✔ | |
170 | H | ✔ | |||||
172 | E2 | 154.31 | 128.947 | 184.763 | 55.816 | ✔ | ✔ |
174 | E2 | 139.91 | 129.548 | 176.8 | 47.252 | ✔ | ✔ |
182 | H | 148.48 | 117.604 | 195.388 | 77.784 | ✔ | |
184 | H | 136.775 | 113.395 | 163.366 | 49.971 | ✔ | |
186 | H | 142.943 | 116.584 | 181.127 | 64.543 | ✔ | |
188 | H | ✔ | |||||
190 | H | 180.28 | 115.232 | 208.892 | 93.66 | ✔ | |
192 | H | 156.778 | 121.067 | 184.443 | 63.376 | ✔ | |
196 | X | 166.715 | 137.05 | 183.725 | 46.675 | ||
198 | X | 161.427 | 137.1 | 223.511 | 86.411 | ||
202 | X | 177.771 | 135.833 | 196.231 | 60.398 | ||
206 | X | 178.27 | 140.421 | 205.967 | 65.546 | ||
212 | X | 183.598 | 141.789 | 217.686 | 75.897 | ||
218 | T2C | 195.333 | 130.635 | 316.825 | 186.19 | ✔ | ✔ |
220 | H | 170.216 | 120.906 | 317.006 | 196.1 | ✔ | |
222 | H | ✔ | |||||
226 | T2D | 188.009 | 150.631 | 229.475 | 78.844 | ✔ | ✔ |
228 | T2D | 133.578 | 118.2 | 152.534 | 34.334 | ✔ | ✔ |
230 | H | 142.084 | 110.959 | 179.786 | 68.827 | ✔ | |
232 | T1A | 141.32 | 115.79 | 155.252 | 39.462 | ✔ | |
234 | H | ✔ | |||||
236 | T1A | ✔ | |||||
238 | T1A | 146.74 | 120.588 | 163.096 | 42.508 | ✔ | |
242 | H | 166.32 | 143.79 | 202.252 | 58.462 | ✔ | |
244 | H | 147.206 | 115.181 | 197.838 | 82.657 | ✔ | |
246 | E2 | ✔ | ✔ | ||||
248 | T2C | 146.941 | 128.79 | 167.802 | 39.012 | ✔ | ✔ |
250 | E2 | ✔ | ✔ | ||||
254 | T2C | 184.612 | 148.452 | 226.275 | 77.823 | ✔ | ✔ |
256 | H | 149.581 | 131.797 | 181.258 | 49.461 | ✔ | |
258 | H | 163.581 | 135.16 | 227.925 | 92.765 | ✔ | |
262 | X | 187.25 | 153.025 | 216.713 | 63.688 | ||
264 | H | ✔ | |||||
266 | X | 184.591 | 145.828 | 200.406 | 54.578 | ||
268 | X | 211.534 | 189.824 | 293.409 | 103.585 | ||
270 | X | 176.493 | 134.692 | 182.49 | 47.798 | ||
274 | H | 165.194 | 152.004 | 197.879 | 45.875 | ✔ | |
276 | H | 207.691 | 154.513 | 268.775 | 114.262 | ✔ | |
280 | H | 196.265 | 146.468 | 236.122 | 89.654 | ✔ | |
282 | H | ✔ | |||||
284 | H | 172.484 | 124.512 | 191.048 | 66.536 | ✔ | |
288 | T1B | 174.823 | 120.912 | 240.868 | 119.956 | ✔ | |
294 | T1B | ✔ | |||||
296 | T1C | 172.266 | 136.523 | 249.397 | 112.874 | ✔ | |
306 | R1B | 135.04 | 111.836 | 175.426 | 63.59 | ✔ | ✔ |
308 | R1B | 145.448 | 130.644 | 196.753 | 66.109 | ✔ | ✔ |
310 | H | ✔ | |||||
312 | H | ✔ | |||||
314 | X | 185.551 | 142.377 | 194.58 | 52.203 | ||
316 | X | 192.964 | 125.025 | 237.809 | 112.784 | ||
320 | X | 190.5 | 119.506 | 270.648 | 151.142 | ||
322 | H | ✔ | |||||
324 | X | 194.728 | 144.047 | 281.585 | 137.538 | ||
328 | X | 175.73 | 149.311 | 326.779 | 177.468 | ||
336 | H | 147.309 | 110.679 | 223.24 | 112.561 | ✔ | |
338 | X | 180.55 | 128.311 | 191.98 | 63.669 | ||
342 | X | ||||||
344 | X | 191.532 | 122.608 | 205.782 | 83.174 | ||
346 | X | 170.863 | 145.521 | 173.785 | 28.264 | ||
350 | X | 178.061 | 128.689 | 196.973 | 68.284 | ||
352 | X | 183.844 | 133.152 | 203.237 | 70.085 | ||
354 | X | ||||||
358 | X | ||||||
360 | X | 184.729 | 117.03 | 190.83 | 73.8 | ||
362 | X | 191.642 | 121.77 | 289.364 | 167.594 | ||
370 | X | 190.976 | 122.748 | 296.334 | 173.586 | ||
372 | H | 169.734 | 113.919 | 212.009 | 191.09 | ✔ | |
374 | T2A | ✔ | ✔ | ||||
376 | T2C | ✔ | ✔ | ||||
380 | X | 193.931 | 118.783 | 216.974 | 98.191 | ||
382 | X | 187.229 | 129.394 | 221.764 | 92.37 | ||
384 | X | 192.489 | 133.18 | 197.468 | 64.288 | ||
386 | X | 187.489 | 123.18 | 207.468 | 84.288 | ||
390 | X | 180.862 | 120.746 | 214.232 | 93.486 | ||
392 | X | 191.623 | 126.211 | 205.629 | 79.418 | ||
406 | X | 176.623 | 123.211 | 225.629 | 102.418 | ||
412 | R1A | 193.426 | 120.306 | 281.224 | 160.918 | ✔ | |
414 | T1B | 182.169 | 113.654 | 227.654 | 114 | ✔ | |
416 | R1A | 170.456 | 112.05 | 219.238 | 107.188 | ✔ | |
420 | X | ||||||
422 | X | ||||||
426 | X | 176.332 | 144.982 | 200.638 | 55.656 | ||
428 | X | 178.467 | 112.391 | 208.232 | 95.841 | ||
432 | X | 172.776 | 119.618 | 192.262 | 72.644 | ||
442 | H | 190.682 | 145.073 | 287.774 | 142.701 | ✔ | |
446 | E2 | ✔ | ✔ | ||||
448 | X | 209.309 | 155.192 | 297.991 | 142.799 | ||
450 | X | 199.723 | 125.112 | 310.076 | 184.964 | ||
464 | H | 160.604 | 119.066 | 193.752 | 74.686 | ✔ | |
466 | E2 | ✔ | ✔ | ||||
470 | H | 143.25 | 116.832 | 166.954 | 50.122 | ✔ | |
472 | E2 | 163.308 | 119.957 | 187.37 | 67.413 | ✔ | ✔ |
474 | H | 163.476 | 121.175 | 171.586 | 50.411 | ✔ | |
476 | T1B | 208.878 | 144.578 | 295.197 | 150.619 | ✔ | |
478 | T1B | ✔ | |||||
481 | H | 189.654 | 122.573 | 290.198 | 167.625 | ✔ | |
485 | T1C | 194.172 | 161.523 | 296.41 | 134.887 | ✔ | |
487 | H | 165.668 | 116.098 | 299.208 | 183.11 | ✔ | |
491 | N3 | ✔ | |||||
493 | T1C | 196.118 | 118.226 | 283.964 | 165.738 | ✔ | |
495 | H | 172.827 | 160.222 | 200.849 | 40.627 | ✔ | |
507 | X | ||||||
526 | H | 181.551 | 123.824 | 219.34 | 95.516 | ✔ | |
528 | R2 | 177.551 | 113.924 | 199.324 | 85.4 | ✔ | |
532 | H | 155.571 | 137.435 | 193.596 | 56.161 | ✔ | |
534 | H | ✔ | |||||
536 | E2 | ✔ | ✔ | ||||
538 | H | 162.539 | 115.185 | 195.999 | 80.814 | ✔ | |
540 | H | 178.268 | 142.035 | 280.391 | 138.356 | ✔ | |
544 | H | 172.569 | 116.159 | 203.385 | 87.226 | ✔ | |
548 | H | ✔ | |||||
554 | H | 198.014 | 112.315 | 205.929 | 93.614 | ✔ | |
556 | T2D | 134.136 | 110.134 | 164.631 | 54.497 | ✔ | ✔ |
558 | E2 | ✔ | ✔ | ||||
560 | H | ✔ | |||||
564 | N3 | 164.302 | 135.567 | 187.228 | 51.661 | ✔ | |
574 | H | ✔ | |||||
578 | H | 145.543 | 114.816 | 185.662 | 70.846 | ✔ | |
582 | H | 145.782 | 112.054 | 188.418 | 76.364 | ✔ | |
584 | H | 171.92 | 118.073 | 198.94 | 80.867 | ✔ | |
586 | H | 188.267 | 120.89 | 192.344 | 71.454 | ✔ | |
588 | H | 142.464 | 127.835 | 163.073 | 35.238 | ✔ | |
592 | R2 | 198.572 | 150.694 | 308.15 | 157.456 | ✔ | |
598 | H | 130.993 | 115.929 | 163.524 | 47.595 | ✔ | |
598 | H | 150.993 | 120.855 | 203.446 | 82.591 | ✔ |
Shaded rows denote entirely creaky speaking turn.*
Appendix C.
Patient A Summary
Mean | Min | Max | Range | |
---|---|---|---|---|
Overall Mean : | 195.298 | 148.785 | 266.402 | 117.617 |
Pain | 194.868 | 149.127 | 267.366 | 118.238 |
Pain: Opioid | 184.919 | 142.710 | 253.420 | 110.709 |
Pain Non-Opioid | 197.316 | 150.706 | 270.797 | 120.090 |
Non-Pain | 202.791 | 142.818 | 249.591 | 106.773 |
Pain: Opioid | Mean | Min | Max | Range |
Narration of pain and description of symptoms | 173.256 | 129.790 | 221.426 | 91.635 |
Request for Opioid treatment | 175.372 | 142.901 | 257.327 | 114.426 |
Positive assessments / satisfaction with opioids. | 200.242 | 145.502 | 259.110 | 113.608 |
Opioid related threats/red flags | 202.320 | 156.212 | 284.349 | 128.137 |
Pain: Non-Opioid | ||||
Unclear request regarding Prescription | 208.608 | 155.528 | 290.200 | 134.672 |
Request for information and logistics | 195.632 | 148.050 | 249.856 | 101.806 |
Negative assessments / dissatisfaction with non-opioid treatments. | 198.157 | 153.377 | 277.530 | 124.153 |
Disagreement or resistance | 201.411 | 157.501 | 283.651 | 126.149 |
Other Pain Related Utterances | 196.490 | 149.827 | 269.928 | 120.101 |
Patient B Summary
Mean | Min | Max | Range | |
---|---|---|---|---|
Overall Mean : | 171.108 | 126.549 | 220.609 | 94.744 |
Pain | 165.816 | 124.725 | 219.141 | 95.416 |
Pain: Opioid | 161.017 | 129.391 | 214.048 | 84.657 |
Pain Non-Opioid | 166.968 | 123.606 | 220.363 | 97.998 |
Non-Pain | 182.552 | 130.493 | 223.785 | 93.291 |
Pain: Opioid | Mean | Min | Max | Range |
Narration of pain and description of symptoms | 152.509 | 126.151 | 182.978 | 56.827 |
Request for Opioid treatment. | 150.993 | 118.655 | 212.042 | 93.386 |
Positive assessments / satisfaction with opioids. | 175.799 | 142.616 | 246.545 | 103.929 |
Expressions of uncertainty or ambiguity about opioids. | 175.629 | 135.959 | 236.967 | 101.008 |
Opioid related threats/red flags | 151.908 | 126.322 | 182.213 | 55.892 |
Pain: Non-Opioid | ||||
Request for Non-Opioid treatment. | 167.327 | 122.899 | 219.862 | 102.014 |
Request for information and logistics. | 184.222 | 133.381 | 235.004 | 101.623 |
Positive assessments /satisfaction with non-opioid treatments. | 154.733 | 112.666 | 216.994 | 104.329 |
Negative assessments / dissatisfaction with non-opioid treatments. | 180.428 | 125.182 | 236.408 | 111.226 |
Expressions of uncertainty or ambiguity about a non-opioid treatment plan. | 180.394 | 134.073 | 260.281 | 126.208 |
Agreement | 164.302 | 135.567 | 187.228 | 51.661 |
Other Pain Related Utterances | 164.747 | 122.899 | 215.644 | 94.436 |
Notes:
While epilaryngeal constriction passively lowers pitch, it does not necessarily constitute active pitch lowering. As Moisik (2013) puts it, the constriction synergizes with low pitch.
It is important to note that epilaryngeal constriction could result in different types of non-modal phonation, including varying degrees of creak like “harsh” and “pure” (Moisik, 2013). In the present study, we collectively use the term “creak” to refer to all of its variants.
Using a Fourier analysis of the first few seconds of the recording, Audacity creates a noise profile used in filtering out the rest of the recording. Both audio files were filtered using the following setting: a noise reduction of 12 decibels with a sensitivity parameter of 3 and frequency smoothing bands set at 0. Audacity can also generate the “noise residue,” or the noise to be filtered out, which was useful in verifying that the audio was not compromised.
In the rare case where all three coders selected distinct codes, each made a case for their decision until an agreement was reached.
Our data also shows that both patients resort to more straightforward demands only when implicit requests are left unaddressed.
References
- Anderson RC, Klofstad CA, Mayew WJ, et al. (2014) Vocal fry may undermine the success of young women in the labor market. PloS one 9(5): e97506. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ball MJ, Perkins MR, Müller N, et al. (2008) The Handbook of Clinical Linguistics. Oxford: Blackwell Publishing Ltd. [Google Scholar]
- Bänziger T and Scherer KR (2005) The role of intonation in emotional expressions. Speech communication 46(3–4): 252–267. [Google Scholar]
- Bendor D and Wang X (2005) The neuronal representation of pitch in primate auditory cortex. Nature 436(7054): 1161. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Berry DA (2001) Mechanisms of modal and nonmodal phonation. Journal of Phonetics 29(4): 431–450. [Google Scholar]
- Boersma P and Weenink D (2013) Praat: doing phonetics by computer [Computer program]. Version 5.3.51. http://www.praat.org/
- Borkowska B and Pawlowski B (2011) Female voice frequency in the context of dominance and attractiveness perception. Animal Behaviour 82(1): 55–59. [Google Scholar]
- Burgess DJ, Crowley-Matoka M, Phelan S, et al. (2008) Patient race and physicians’ decisions to prescribe opioids for chronic low back pain. Social science & medicine 67(11): 1852–1860. [DOI] [PubMed] [Google Scholar]
- Cernak M, Orozco-Arroyave JR, Rudzicz F, et al. (2017) Characterisation of voice quality of Parkinson’s disease using differential phonological posterior features. Computer Speech & Language 46: 196–208. [Google Scholar]
- Chafe WL (1993) Prosodic and functional units of Language In: Lampert MD and Edwards JA (eds) Talking Data: Transcription and Coding in Discourse Research. Hillsdale, N.J.: Lawrence Erlbaum Associates, pp. 24–43. [Google Scholar]
- Couper-Kuhlen E (2015) Intonation and discourse In: Tannen D, Hamilton HE, and Schiffrin D (eds) The Handbook of Discourse Analysis. John Wiley & Sons, Inc; Hoboken, NJ, USA, pp. 82–104. [Google Scholar]
- Cruttenden A (1997) Intonation. Cambridge: Cambridge University Press. [Google Scholar]
- Dudy S, Bedrick S, Asgari M, et al. (2018) Automatic analysis of pronunciations for children with speech sound disorders. Computer speech & language 50: 62–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Esling JH (1978) The identification of features of voice quality in social groups. Journal of the International Phonetic Association 8(1–2): 18–23. [Google Scholar]
- Esling JH (2012) Voice Quality In: Chapelle CA (ed.) The Encyclopedia of Applied Linguistics. Oxford, UK: Blackwell Publishing Ltd; DOI: 10.1002/9781405198431.wbeal1271. [DOI] [Google Scholar]
- Esling JH, Moisik SR, Benner A, et al. (2019) Voice Quality: The Laryngeal Articulator Model Cambridge Studies in Linguistics ; 162 Cambridge, United Kingdom: ; Cambridge University Press. [Google Scholar]
- Fleiss JL (1971) Measuring nominal scale agreement among many raters. Psychological bulletin 76(5): 378–382. [Google Scholar]
- Gerhard D (2003) Pitch Extraction and Fundamental Frequency: History and Current Techniques. Department of Computer Science, University of Regina, Regina. [Google Scholar]
- Gill VT, Halkowski T and Roberts F (2001) Accomplishing a request without making one: A single case analysis of a primary care visit. Text–Interdisciplinary Journal for the Study of Discourse 21(1–2): 55–81. [Google Scholar]
- Gordon M and Ladefoged P (2001) Phonation types: a cross-linguistic overview. Journal of phonetics 29(4): 383–406. [Google Scholar]
- Grivicic T and Nilep C (2004) When phonation matters: The use and function of yeah and creaky voice.: 1–11.
- Gumperz JJ (1982) Discourse Strategies. Cambridge: Cambridge University Press. [Google Scholar]
- Heath C (2002) Demonstrative suffering: the gestural (re) embodiment of symptoms. Journal of Communication 52(3): 597–616. [Google Scholar]
- Henry SG and Eggly S (2013) The effect of discussing pain on patient-physician communication in a low-income, black, primary care patient population. The Journal of Pain 14(7): 759–766. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Henry SG, Chen M, Matthias MS, et al. (2016) Development of the Chronic Pain Coding System (CPCS) for characterizing patient-clinician discussions about chronic pain and opioids. Pain Medicine 17(10): 1892–1905. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Henry SG, Bell RA, Fenton JJ, et al. (2017) Goals of Chronic Pain Management. The Clinical journal of pain 33(11): 955–961. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Henton CG (1986) Creak as a sociophonetic marker. The Journal of the Acoustical Society of America 80(S1): S50–S50. [Google Scholar]
- Højsted J and Sjøgren P (2007) Addiction to opioids in chronic pain patients: a literature review. European Journal of Pain 11(5): 490–518. [DOI] [PubMed] [Google Scholar]
- Hughes HK, Korthuis PT, Saha S, et al. (2015) A mixed methods study of patient–provider communication about opioid analgesics. Patient education and counseling 98(4): 453–461. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Irvine J (2001) ‘Style’ as distinctiveness: The culture and ideology of linguistic differentiation In: Eckert P and Rickford JR (eds) Style and Sociolinguistic Variation. Cambridge, U.K.: Cambridge University Press, pp. 21–43. [Google Scholar]
- Keating P, Garellek M and Kreiman J (2015) Acoustic properties of different kinds of creaky voice. In: Proceedings of the 18th International Congress of Phonetic Sciences, 2015, pp. 0821–1. [Google Scholar]
- Matthias MS, Krebs EE, Collins LA, et al. (2013) “I’m Not Abusing or Anything”: Patient–physician communication about opioid treatment in chronic pain. Patient education and counseling 93(2): 197–202. [DOI] [PubMed] [Google Scholar]
- Matthias MS, Krebs EE, Bergman AA, et al. (2014) Communicating about opioids for chronic pain: A qualitative study of patient attributions and the influence of the patient–physician relationship. European journal of pain 18(6): 835–843. [DOI] [PubMed] [Google Scholar]
- Mendoza-Denton N (2011) The semiotic hitchhiker’s guide to creaky voice: circulation and gendered hardcore in a Chicana/o gang persona. Journal of Linguistic Anthropology 21(2): 261–280. [Google Scholar]
- Merrill JO, Rhodes LA, Deyo RA, et al. (2002) Mutual mistrust in the medical care of drug users. Journal of General Internal Medicine 17(5): 327–333. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Moisik S (2013) The epilarynx in speech Doctoral Dissertation. University of Victoria, Victoria, BC, Canada. [Google Scholar]
- Podesva RJ (2007) Phonation type as a stylistic variable: The use of falsetto in constructing a persona 1. Journal of sociolinguistics 11(4): 478–504. [Google Scholar]
- Roberts F and Kramer JS (2014) Medication and morality: Analysis of medical visits to address chronic pain In: Hamilton H and Chou WS (eds) The Routledge Handbook of Language and Health Communication. London: Rouletge, pp. 477–489. [Google Scholar]
- Robinson JD (2001) Asymmetry in action: Sequential resources in the negotiation of a prescription request. Text–Interdisciplinary Journal for the Study of Discourse 21(1–2): 19–54. [Google Scholar]
- Schilling-Estes N (1998) Investigating “self-conscious” speech: The performance register in Ocracoke English. Language in society 27(1): 53–83. [Google Scholar]
- Slobe T (2018) Style, stance, and social meaning in mock white girl. Language in Society 47(4): 541–567. [Google Scholar]
- Sullivan M and Ferrell B (2005) Ethical challenges in the management of chronic nonmalignant pain: negotiating through the cloud of doubt. The Journal of Pain 6(1): 2–9. [DOI] [PubMed] [Google Scholar]
- Tannen D and Wallat C (1987) Interactive frames and knowledge schemas in interaction: Examples from a medical examination/interview. Social psychology quarterly 50(2): 205–216. [Google Scholar]
- Turk DC and Okifuji A (1997) What factors affect physicians’ decisions to prescribe opioids for chronic noncancer pain patients? The Clinical journal of pain 13(4): 330–336. [DOI] [PubMed] [Google Scholar]
- Van Dijk TA (1999) Critical discourse analysis and conversation analysis. Discourse & Society 10(4): 459–460. [Google Scholar]
- Van Dijk TA (ed.) (2011) Discourse as Structure and Process Reprinted. Discourse studies a multidisciplinary introduction / ed. by van Dijk Teun A.; Vol. 1 London: SAGE. [Google Scholar]
- Wilce JM (1997) Discourse, power, and the diagnosis of weakness: encountering practitioners in Bangladesh. Medical Anthropology Quarterly 11(3): 352–374. [DOI] [PubMed] [Google Scholar]
- Yuasa IP (2010) Creaky voice: A new feminine voice quality for young urban-oriented upwardly mobile American women? American Speech 85(3): 315–337. [Google Scholar]
- Zetterholm E (1998) Prosody and voice quality in the expression of emotions. In: Fifth International Conference on Spoken Language Processing, 1998. [Google Scholar]
- Zimman L (2012) Voices in Transition: Testosterone, Transmasculinity, and the Gendered Voice among Female-to-MaleTransgender People Linguistics Graduate Theses & Dissertations. University of Colorado at Boulder, Colorado. [Google Scholar]
- Zimman L (2013) Hegemonic masculinity and the variability of gay-sounding speech: The perceived sexuality of transgender men. Journal of Language and Sexuality 2(1): 1–39. [Google Scholar]