1. Introduction
Pain is notoriously difficult to verbalize (Lascaratou, Reference Lascaratou2007; Scarry, Reference Scarry1985; Semino, Reference Semino2010). The Japanese language has two major solutions to this problem: metaphors and ideophones.Footnote 1 As in many other languages, conventional pain metaphors and similes in Japanese represent different types of pain primarily using verbs and adjectives, as in sasu-yoona itami ‘stabbing pain’, denki-ga hashiru-yoona itami ‘pain like an electric current running through one’s body’ and surudoi itami ‘sharp pain’ (Kövecses, Reference Kövecses, Lascaratou, Despotopoulou and Ifantidou2008). Ideophones (also known as ‘mimetics’ or ‘expressives’), including onomatopoeia (e.g., clip-clop, woof), are conventional (and controversially, spontaneously created) words that vividly represent various types of sensory experience using speech sound imitatively (Akita & Dingemanse, Reference Akita, Dingemanse and Aronoff2019; Dingemanse, Reference Dingemanse, Akita and Pardeshi2019). Japanese patients often describe their bodily feelings with ideophones, such as kirikiri ‘having a splitting ache’ and shikushiku ‘having a dull pain in one’s stomach’ (Pfizer Japan Inc., 2016; Ueda, Reference Ueda2015), but some manuals for healthcare workers (e.g., Yamauchi, Reference Yamauchi2005) explicit say that these words should be avoided in medical communication because they are too subjective to represent the patient’s pain perception accurately. However, linguists may expect the direct, highly specific semantic representation of ideophones (Akita, Reference Akita2012; Diffloth, Reference Diffloth1972; Dingemanse, Reference Dingemanse2011; Kita, Reference Kita1997) to rather facilitate health communication when used properly.
This study is the first to investigate how reliable ideophones may be as pain descriptions, as compared with nonideophonic metaphors. Section 2 summarizes previous psycholinguistic studies on Japanese pain descriptions. Sections 3 and 4 report on two experiments in which Japanese speakers rated pain expressions on 15 semantic-differential scales. Based on the findings, in Section 5, we argue that ideophonic pain expressions can be more stable but more flexible than metaphorical ones and are worth serious attention in health communication.
2. Pain ideophones in Japanese
Numerous languages of the world, from Niger–Congo to Austroasiatic, Austronesian, Dravidian, and Quechuan languages, have large inventories of ideophones (Akita & Pardeshi, Reference Akita and Pardeshi2019; Dingemanse, Reference Dingemanse2018; Voeltz & Kilian-Hatz, Reference Voeltz and Kilian-Hatz2001). They represent not only sound but also motion (e.g., txingin-txingin ‘hopping’ in Basque; Ibarretxe-Antuñano, Reference Ibarretxe-Antuñano2006), texture (e.g., jobo-jobo ‘sticky, oily on the skin’ in Mundari; Badenoch, 2021), taste (e.g., hóyíhóyíhóyí ‘bitter sharp taste’ in Ewe; Ameka, Reference Ameka, Voeltz and Kilian-Hatz2001), smell (e.g., kpiini-kpiini ‘strongly offensive to sense of smell, often used with reference to flatulence’ in Kisi; Childs, Reference Childs1988), emotion (e.g., kɛ́lá kɛ́lá ‘happy’ in Gbeya; Samarin, Reference Samarin1970), and interoception, including nociception (e.g., ttakkumttakkum ‘piercingly painful’ in Korean; Sohn, Reference Sohn1994) (see also McLean, Reference McLean2021 for a recent discussion on the semantic typology of ideophones). Ideophones are ‘sound-symbolic’ in the sense that they consist of sounds that native speakers interpret as ‘matching’ their meanings and, as such, are discussed as a word-level instance of iconicity (i.e., perceived resemblance between form and meaning) in language (Akita & Dingemanse, Reference Akita, Dingemanse and Aronoff2019). For example, to native speakers of Japanese, voiced obstruents (e.g., /b, d, ɡ, z, ʑ/) systematically sound larger, stronger, dirtier, and less pleasant than their voiceless counterparts (e.g., /p, t, k, s, ɕ/), as illustrated by ideophone pairs such as bokoboko ‘hitting hard’ versus pokopoko ‘hitting lightly’ and zarazara ‘rough’ versus sarasara ‘dry and smooth’ (Hamano, Reference Hamano1998).Footnote 2 , Footnote 3
Japanese speakers think that ideophones allow them to express their subjective internal sensations directly (Ishidate, Reference Ishidate2016; Moriyama, Reference Moriyama1994; Tanaka et al., Reference Tanaka, Okajima and Komatsu2015; Ueda, Reference Ueda2015; see Osaka et al., Reference Osaka, Osaka, Morishita, Kondo and Fukuyama2004 for a neural correlate of pain ideophones; see also Dingemanse, Reference Dingemanse2011, Ch. 11 for an observation that ideophones index epistemic authority, an individual who has privileged access to the conversational topic (Hayano, Reference Hayano, Stivers, Mondada and Steensig2011)). According to Pfizer Japan, Inc.’s (2016) nationwide survey, 80% of patients with chronic throbbing pain reported that they often or sometimes use ideophones to describe their pain. Many of them also reported that when they successfully communicate their pain to medical workers with ideophones, they get a sense of being understood.
Another potential advantage of ideophonic pain expressions is their formal flexibility. Due to their iconic nature, they allow us to adjust our pain descriptions on various dimensions using prosodic and morphological features (Dingemanse, Reference Dingemanse2015; Dingemanse & Akita, Reference Dingemanse and Akita2017). For example, the ideophone biriQ ‘having a momentary electric shock’ can be prolonged (biriiiQ, biriiiiiQ, etc.) to express the duration or intensity of the shock, often accompanied by marked voice quality, such as harsh, strained, or creaky voice and falsetto (see Akita, Reference Akita2021).Footnote 4 Partial reduplication (biribiriQ, biribiribiriQ, etc.) and other types of expressive morphology may also contribute to the aspectual or emphatic dimension of ideophone semantics. This formal flexibility may help patients to express their nuanced feelings that are otherwise difficult to share with others.
Despite these expected benefits, little has been investigated about how effective ideophones may be in the healthcare context (but see Sakamoto et al., Reference Sakamoto, Ueda, Doizaki and Shimizu2014 for an attempt to quantify Japanese speakers’ sound–meaning associations in pain ideophones). The purpose of this study is threefold: 1) to clarify how stable (i.e., shared between patients) people’s understanding of pain ideophones is as compared with that of pain metaphors, 2) to identify which aspects of their meanings tend to be (un)stable, and 3) to examine whether people share the minute semantic differences between prosodic/morphological variants of pain ideophones. Investigating these issues will facilitate our understanding as to whether and how to use ideophones in health communication.
3. Experiment 1
3.1. Method
3.1.1. Participants
We recruited 151 native speakers of Japanese who are self-reportedly not proficient in another language (female: 95, male: 53, other/prefer not to answer: 3; age: 19–75, M = 39.19, SD = 10.68) via CrowdWorks, a Japanese equivalent of Amazon Mechanical Turk. They gave online informed consent and were paid 350 JPY for their participation.
3.1.2. Stimuli
A total of 50 conventional pain expressions in Japanese were used (Table 1). We collected 34 pain ideophones from Iwasaki et al. (Reference Iwasaki, Vinson, Vigliocco and Minami2007a), Akita (Reference Akita2010), Hattori and Higashiyama (Reference Hattori and Higashiyama2010), and Ono (Reference Ono2007) and 16 nonideophonic metaphorical expressions of pain from Kusumi et al. (Reference Kusumi, Nakamoto and Koyasu2010), Maruo et al. (Reference Maruo, Nakae, Maeda, Takahashi-Narita, Morris, Yokoe, Matsuzaki, Shibata and Saitoh2013), and Sakamoto (Reference Sakamoto2019). This stimulus set covers all conventional pain ideophones and representative pain metaphors. Both pain ideophones and metaphors can be newly created, but the latter form a particularly fluid category, as the simile construction potentially gives rise to an unlimited number of expressions.
Table 1. Stimulus pain expressions used in Experiment 1
3.1.3. Procedure
The experiment was conducted online using Google Forms. The participants were asked to rate in a quiet place each of the pain expressions on 15 six-point semantic-differential scales (from 1 to 6), all of which except ‘unpleasantness’ (unpleasant–not unpleasant) were taken from Iwasaki et al.’s (Reference Iwasaki, Vinson, Vigliocco and Minami2007a) similar experiment with English speakers: ‘stimulation’ (stimulating–not stimulating), ‘intensity’ (strong–weak), ‘sharpness’ (sharp–dull), ‘locality’ (local–widespread), ‘momentariness’ (momentary–continuous), ‘achingness’ (aching–not aching), ‘annoyance’ (annoying–not annoying), ‘stabbingness’ (stabbing–not stabbing), ‘surfaceness’ (surface–internal), ‘numbness’ (numb–not numb), ‘burningness’ (burning–not burning), ‘pressure’ (pressing–not pressing), ‘tearingness’ (tearing–not tearing) and ‘pulsation’ (pulsating–not pulsating). According to the McGill Pain Questionnaire (Melzack, Reference Melzack1975), these scales can be classified into ‘sensory’ (sharpness, numbness, achingness, stimulation, burningness, stabbingness, pressure, tearingness), ‘temporal’ (momentariness, pulsation) and ‘evaluative’ (intensity, annoyance, unpleasantness), with the two scales (locality, surfaceness) that we call ‘locational’ left unclassified. The 750 questions (50 expressions x 15 scales) were randomized and divided into 15 sets of 50 questions, one of which each participant answered. The questions were presented in a random order. Each set was answered by 10 or 11 participants, each of whom answered only one set.
The current experiment essentially took an exploratory approach, as little has been known about the semantic stability of ideophones and metaphors as pain descriptors. However, the idea that pain ideophones are highly subjective would predict that they receive more varied ratings than nonideophonic metaphors.
All statistical analysis was conducted in R version 4.3.1 (R Core Team, 2023). Cumulative link mixed models were constructed using the ordinal package (Christensen, Reference Christensen2019).
3.2. Results
3.2.1. Semantic stability of ideophones versus metaphors
The experiment revealed that the ratings of pain ideophones are overall as stable as those of pain metaphors. Figure 1 shows the standard deviations of the ratings for all expressions.
Figure 1. Standard deviations of the ratings of the 50 pain expressions, from the lowest to the highest (the blue dots represent the means). Ideophones were generally rated as stable as metaphors.
The ratings were most stable for the ideophone piripiri ‘irritated continuously on the skin’ (M = 1.07, SD = 0.37), followed by the metaphors denki-ga hashiru-yoona itami ‘pain like an electric current running through one’s body’ (M = 1.08, SD = 0.57) and sasu-yoona itami ‘stabbing pain’ (M = 1.09, SD = 0.67). A linear model predicting SD from expression type (ideophones versus metaphors) yielded no statistically reliable effect of the predictor (b = 0.02, SE = 0.03, t = 0.63, p = .53), indicating that ideophones are as objective and stable as nonideophonic metaphors. However, it should be noted that some of both ideophones and metaphors, especially less conventional ones, such as igaiga ‘scratchy on the throat’ (M = 1.53, SD = 0.25), zuun ‘having a dull, heavy pain deep inside the body’ (M = 1.41, SD = 0.29), wareru-yoona itami ‘cracking pain’ (M = 1.39, SD = 0.55) and shiborareru-yoona itami ‘squeezing pain’ (M = 1.39, SD = 0.49), did receive relatively varied ratings.Footnote 5
Another parallelism between ideophones and metaphors was found for the stability of scales. Figure 2 shows the standard deviations of the ratings as a function of the 15 scales. In both ideophones and metaphors, intensity and unpleasantness tended to be rated more consistently across participants, whereas pulsation and tearingness ratings tended to vary more widely.
Figure 2. Standard deviations of the ratings of pain ideophones and metaphors on the 15 scales, from the lowest to the highest in ideophones (the blue dots represent the means). The two types of expressions were found to be similar to the stability of the scales.
When the scales were grouped into four according to the McGill Pain Questionnaire (Figure 3), a reliable difference between ideophones and metaphors was obtained for the evaluative scales (i.e., intensity, annoyance, unpleasantness). A linear model predicting standard deviation from expression type, scale type and the interaction of the two variables, with the sensory scales as a reference level, revealed a reliable interaction (b = −0.39, SE = 0.08, t = −4.57, p < .001). Specifically, although both pain ideophones and metaphors were rated most stably on the evaluative scales, this difference in stability was larger for metaphors than for ideophones.Footnote 6 , Footnote 7
Figure 3. Standard deviations of the ratings of pain ideophones and metaphors on the four types of scales (the blue dots represent the means). Both types of expressions are most stable on the evaluative scales, but the stability of metaphors is particularly high on these scales.
3.2.2. Interim conclusion
The current results suggest that, overall, ideophones are as stable as metaphors in Japanese pain descriptions. This finding might be surprising to those who assume ideophones to be too vague and subjective. They are words with specific meanings and systematic sound symbolism, and Japanese speakers share them at least to a comparable degree with metaphors. Both pain ideophones and pain metaphors are ‘figurative’ in that they represent pain in terms of a concrete concept (i.e., speech sound and physical action). Japanese speakers have a rather clear understanding of these linguistic conventions and use them in describing their personal experience.
However, it should be noted that the current results might instead mean that pain ideophones are as unstable as pain metaphors, as there is no previously established baseline and we just compared these two types of expressions. Moreover, what is crucially missing in the current experiment is the context. The 50 pain expressions were presented in isolation, without any syntactic or semantic specification. This can be especially problematic for polysemous ideophones. Unlike the pain metaphors, which are headed by the noun itami ‘pain’, some of the pain ideophones have non-nociceptive, often psychological meanings, too (e.g., jiin ‘having a heavy, numb feeling; deeply touched’, piripiri ‘irritated continuously on the skin; uptight’, shikushiku ‘having a dull pain in one’s stomach; weeping’). Furthermore, we did not provide prosodic information, which would make a considerable contribution to the meaning of ideophones (Akita, Reference Akita2021; Dingemanse & Akita, Reference Dingemanse and Akita2017). To overcome these limitations, we conducted Experiment 2, using more natural stimuli.
4. Experiment 2
4.1. Method
4.1.1. Participants
One hundred eighty Japanese monolinguals (female: 113, male: 66, other/prefer not to answer: 1; age: 19–77, M = 39.89, SD = 10.93) were recruited via CrowdWorks, gave online informed consent and were paid 350 JPY for their participation. Forty-one of them (22.78%) also participated in Experiment 1 half a year before. All the results presented below were replicated with the subset of the data that does not include these overlapping participants (for details, see the R script provided on the project’s OSF page).
4.1.2. Stimuli
Six Q-ending pain ideophones, which are most flexible in prosody and morphology (Akita, Reference Akita2020), and six pain metaphors were selected from the stimulus set in Experiment 1. These expressions were presented auditorily in simple sentence frames that consist of a body-part noun and a past-tense verb (the light verb shi-ta ‘did’ for ideophones and the verb itan-da ‘hurt’ for metaphors), as in Table 2. The metaphors in Experiment 1 took the form of [adnominal + N] (e.g., uzuku-yoona itami ‘throbbing pain’), which was converted into [adverbial + V] (e.g., uzuku-yooni itan-da ‘had a throbbing pain’) in Experiment 2, fitted into the sentence frame.
Table 2. Stimulus pain expressions used in Experiment 2
These pain expressions were presented with or without formal modifications. Modifications were either lexical (addition of the degree adverb sukoshi ‘a little bit’ or kanari ‘fairly’), morphological (partial reduplication; e.g., biriQ ‘having a momentary electric shock’ > biribiriQ), or prosodic (vowel lengthening with strained voice, distinguishing plain (e.g., biriQ), strong (e.g., biriiiQ) and extrastrong prosody (e.g., biriiiiiQ)). We did not modify the morphology and prosody of metaphors, as partial reduplication is not possible with them (e.g., uzuku-yooni ‘throbbing’ > *uzuuzuku-yooni) and prosodic modification also often makes them sound unnatural (e.g.,??uzuuuku-yooni). The semantic relevance of expressive morphology and prosody in ideophones discussed in the literature (Akita, Reference Akita2021; Dingemanse & Akita, Reference Dingemanse and Akita2017) motivates us to pay special attention to whether Japanese speakers share the minute semantic differences the morphological and prosodic modifications of ideophones may make.
4.1.3. Procedure
The procedure of the rating task was the same as Experiment 1, except that the stimuli were presented auditorily. Participants were instructed to listen to them as many times as they liked wearing headphones or earphones in a quiet place. A total of 18 sets of 45 randomized stimuli and their counterbalanced version were made, and half of the participants answered the former and the other half the latter. Each participant answered only one of them.
This experiment is again exploratory in nature. However, if the sentence frames get rid of the polysemy issue raised in Experiment 1, it is expected that pain ideophones can exhibit greater interrater stability than pain metaphors.
4.2. Results
The results revealed that, when presented in sentence frames, ideophones are generally understood more consistently than metaphors. Figure 4 shows the standard deviations of the ratings of 12 pain expressions without formal modifications. A linear model that predicts standard deviation from expression type (ideophones versus metaphors) revealed that ideophone ratings are generally more stable than metaphor ratings (b = −0.16, SE = 0.06, t = −2.90, p < .01). This difference was not obtained when we reanalyzed the results of Experiment 1 by limiting ourselves to the 12 pain expressions used in the current experiment (b = 0.06, SE = 0.07, t = 0.84, p = .41), indicating the importance of sentential context in ideophone semantics.
Figure 4. Standard deviations of the ratings of the 12 pain expressions without formal modifications, from the lowest to the highest (the blue dots represent the means). Ideophones were generally rated more consistently across participants.
Similar tendencies were obtained across the four scale types. As shown in Figure 5, both pain ideophones and metaphors again exhibited highest stability on the evaluative scales. A linear model predicting standard deviation from expression type, scale type and their interaction, with the sensory scales as the reference level, revealed that unlike in Experiment 1, the stability of metaphor ratings on the evaluative scales was not particularly striking compared with that of ideophone ratings (b = −0.13, SE = 0.14, t = −0.94, p = .35).
Figure 5. Standard deviations of the ratings of pain ideophones and metaphors in sentence frames on the four types of scales (the blue dots represent the means). Both ideophones and metaphors were rated most stably on the evaluative scales. Metaphors’ stability was not particularly high compared to ideophones’ unlike in Experiment 1.
Moreover, as shown in Figure 6, the 15-scale ratings of ideophones were equally stable across different morphological/prosodic forms. A linear model predicting standard deviation from formal modification revealed no reliable difference between plain (i.e., without modification; e.g., biriQ) and strong (e.g., biriiiQ; b = −0.03, SE = 0.06, t = −0.50, p = .62), extra strong (e.g., biriiiiiQ; b = 0.03, SE = 0.06, t = 0.58, p = .56) or reduplicated forms (e.g., biribiriQ; b = −0.05, SE = 0.06, t = −0.83, p = .41).Footnote 8
Figure 6. Standard deviations of the ratings of the six pain ideophones with and without formal modifications (the blue dots represent the means). All prosodic/morphological variants of the ideophones were rated stably.
To give an example, as shown in Figure 7, the participants shared the sound-symbolic intuition that ideophones with plain, strong and extra strong forms increasingly represent more intense pain. They rated plain forms weaker than strong forms (b = −3.21, SE = 0.57, z = −5.66, p < .001), which were in turn rated weaker than extra strong forms (b = −2.27, SE = 0.56, z = −4.08, p < .001). The rated intensity of plain and partially reduplicated forms was not reliably different (b = −0.83, SE = 0.46, z = −1.83, p = .07).
Figure 7. Intensity ratings of pain ideophones with and without formal modifications, from weak (1) to strong (6) (the blue dots represent the means). Ideophones with extra strong prosody were rated stronger than those with strong prosody, which were in turn rated stronger than plain and reduplicated ideophones.
Another example is given in Figure 8, where partially reduplicated forms were rated more continuous than plain (b = −1.72, SE = 0.46, z = −3.75, p < .001), strong (b = −1.55, SE = 0.41, z = −3.76, p < .001) and extra strong forms (b = −1.03, SE = 0.41, z = −2.54, p < .05).
Figure 8. Momentariness ratings of pain ideophones with and without formal modifications, from momentary (1) to continuous (6) (the blue dots represent the means). Reduplicated ideophones were rated more continuous than plain, strong and extra strong ideophones.
5. Discussion and conclusion
Ideophones pack detailed sensory information in a limited number of speech sounds and allow us to communicate our subjective perceptions. The purpose of the current study was to examine how reliable ideophones are in comparison with metaphors in Japanese pain descriptions. Contrary to what some medical workers assume, our experiments on demonstrated that, when presented in proper context, conventional pain ideophones are generally understood more stably (or less unstably) than conventional pain metaphors. It was also shown that even fine-tuned sound-symbolic intuitions, such as the intensity of emphatic prosody and the continuousness of reduplication, are widely shared across individuals. It is also worth noting that both ideophones and metaphors are most stable in the evaluative dimension of pain perception.
The current study points to the practical applicability of ideophones or, more broadly, linguistic iconicity. Iconicity is now attracting renewed attention in cognitive science (Imai & Kita, Reference Imai and Kita2014; Perniss et al., Reference Perniss, Thompson and Vigliocco2010). However, this trend centers on language per se, especially on its acquisition and evolution, and little has been explored as to how useful they can be in our daily lives (for a related review, see Nielsen & Dingemanse, Reference Nielsen and Dingemanse2021). Our findings suggest that ideophones may facilitate not only children’s language acquisition and the evolution of human language but also our everyday communication. This project’s OSF page provides the semantic-differential profile of each pain expression obtained in the two experiments, which we hope will help medical workers to understand patients’ pain descriptions in actual clinical interactions. The data might also be useful for L2 learners of Japanese who do not have clear intuitions about what those ideophones mean.
The current study has some important limitations. First, our experiments, which showed that pain ideophones are in general more stable than pain metaphors, do not guarantee that pain ideophones are stable and should be used extensively in medical communication. It might be that our results are attributed to the highly unstable meaning of pain metaphors, many of which are hyperboles that refer to unreal events, such as one’s body being split or squeezed. Since people normally have never been split or squeezed—although they probably have some experience of parts of their body (e.g., a finger) being split or squeezed—they would not be sure what a splitting/squeezing pain is like. Therefore, a separate benchmark would be necessary to evaluate the real applicability of the current findings. Second, this study fails to pinpoint exactly what factor made ideophone ratings more stable than metaphors in Experiment 2. The two experiments differed as to whether pain expressions were placed in sentence frames, whether they were presented visually or auditorily, and whether they varied in prosody and morphology. All these conditions converge in contributing to the natural context of ideophone use, but future research needs to disentangle the effects of these factors.Footnote 9 Third, we did not collect the participants’ detailed personal information, such as their medical history and profession. These pieces of information may help us to see what doctors, nurses, and patients know about ideophones and metaphors and how they use them.
Despite these limitations, it is hoped that future research will extensively connect these experimental findings with clinical observations. Do ideophones help doctors to identify particular symptoms or even diagnose particular diseases? Should doctors and nurses use ideophones themselves, or should they just repeat or paraphrase patients’ ideophones to express understanding and sympathy (Ueda, Reference Ueda2015)? Are dialectal ideophones and newly coined ideophones also understood stably across patients? How do ideophones interact with gestures, which often contribute to ideophones’ iconic representation (Dingemanse, Reference Dingemanse2011; Dingemanse & Akita, Reference Dingemanse and Akita2017; Kita, Reference Kita1997), in pain descriptions (Rowbotham et al., 2014)? It is also hoped that interactional linguists and conversation analysts will analyze actual doctor–patient interactions and make practical suggestions about the usage of pain ideophones and metaphors. Furthermore, in light of the semantic typology of ideophones, it will be essential to extend the current project to other semantic domains, such as haptic and bodily feelings other than pain, and to other languages (McLean, Reference McLean2021; Nielsen & Dingemanse, Reference Nielsen and Dingemanse2021).
Data availability
The experimental instructions, data, additional figures and R code are all available on the project’s OSF page: https://osf.io/hc8b4/.
Acknowledgments
I would like to thank Mark Dingemanse and another reviewer of Language and Cognition for their insightful comments. I also got useful feedback from Hinano Iida and Ryota Kido.
Funding
This work was supported by JSPS KAKENHI Grant Number 23H00627.
Competing interest
The author has no competing interests to declare.
Open access
