Bilingualism

Miroslava Albiter, Caitlin Morlett, Renee Ma, Jaquelin Trujillo, and Zirui(Ray)

Have you ever heard your friend or family speak two languages in one phrase? Have you ever spoken two languages in a sentence? We look deeper into how code-switching affects phonological convergence, specifically in Spanish-English bilinguals. A key phonological difference between English and Spanish is the articulation of word-initial voiceless stops such as /p/, /t/, and /k/. Therefore, we specifically analyzed how the Voiced Time Onset (VOT) measures of Spanish-English bilinguals are affected by code-switching between Spanish and English. Sixteen English-Spanish bilinguals were recorded and asked to read aloud the Rainbow passage, a passage with English sentences, Spanish sentences, and code-switched English-Spanish sentences. PRAAT was used to measure the participants’ VOTs to compare the differences to a baseline VOT measure of monolingual English and Spanish speakers (Castañeda Vicente, 1986; Lisker & Abramson, 1964).

After data collection and analysis, we discovered that both VOTs of English and Spanish were lengthened during code-switching, albeit for different reasons. As Spanish VOT extended due to phonological convergence, English VOT also unexpectedly extended. We observed evidence for hyper-articulation, which can further explain our conclusion. However, limitations are reckoned with, and thus, fields of phonological change within code-switched contexts are explored.

[expander_maker id=”1″ more=”Read more” less=”Read less”]

Introduction and Background

In this article, we investigated code-switching in English-Spanish bilingual college students, focusing on their phonological change along with the process of code-switching through Voice Onset Time. Code-switching “is a practice common among bilinguals whereby speakers use both languages in a single utterance” (Balukas & Koops, 2015). We will analyze whether English-Spanish bilingual college students demonstrate phonological convergence of VOT in code-switching sentences. The age range of the 16 tested participants varies, with 14 being 20 years old and the other two between 40 and 50 years old. After collecting data, we examined the Voice Onset Time (VOT) in different circumstances and the subsequent phonological convergence in English-Spanish code-switched sentences.

According to previous research, English monolingual speakers typically produce a longer VOT of word-initial voiceless stops than Spanish monolingual speakers (Balukas & Koops, 2015). Samples from bilingual participants were collected and analyzed using Praat, with a particular focus on VOT changes and the evidence of subsequent phonological convergence. As such, we are investigating a potential phenomenon wherein the English VOT is observed to shorten while the Spanish VOT lengthens in English-Spanish code-switched sentences. This dynamic interaction may be further interpreted as evidence of phonological convergence, reflecting an underlying adjustment mechanism within bilingual speech production systems. Due to the lack of ready access to true monolingual speakers in the LA area, we will use data from the BYU Scholars Archive to compare the VOT of our bilingual speakers to the average VOT for /p/, /t/, and /k/ in both English and Spanish. In the BYU Scholars Archive, the Spanish baseline was recorded at the University of Barcelona in 1986 and was taken from 10 monolingual Spanish speakers (Castañeda Vicente, 1986). The English baseline in the archive was recorded at the University of Pennsylvania in 1964 based on 4 American monolingual English speakers (Lisker & Abramson, 1964).

Figure 1 – Chart from BYU Scholars Archive showing the average VOT is ms for English bilinguals and Spanish bilinguals

Phonological properties, including VOTs, can be differentiated across various dialects, even within the same language. In this research, our participants predominantly speak LAVS (Los Angeles Vernacular Spanish), while three speak the Castellano dialect.

In this study, we will investigate whether code-switching influences VOT in English-Spanish bilingual speakers. Testing the specific phonological changes in VOTs in English and Spanish during code-switching. It finally aims to detect the phonological convergence during code-switching in English-Spanish bilinguals. Our hypotheses of VOT and phonological convergence are as follows:

1. English-Spanish bilingual speakers will shorten English VOT and lengthen Spanish VOT compared to true monolingual speakers of each respective language.
2. If the speaker’s L1 is English and their L2 is Spanish, we expect the VOT for Spanish /p/, /t/, and /k/ to be longer than Spanish monolingual speakers. We also expect their English VOT for /p/, /t/, and /k/ to reflect the VOT of monolingual English speakers.
3. If the speaker’s L1 is Spanish and their L2 is English, we expect the VOT for English /p/, /t/, and /k/ to be shorter than English monolingual speakers. We would also expect their Spanish VOT for /p/, /t/, and /k/ to reflect the VOT of monolingual Spanish speakers.

If a participant has been discouraged from speaking Spanglish, they will be more likely to have longer VOT in Spanish despite differences in their L1 or L2.

To clarify related concepts, VOT is “the temporal relationship between laryngeal pulsing and the onset of consonant release” (Molfese & Narter, 1997). Phonological convergence, on the other hand, refers to the process by which speakers in a communicative interaction adjust their phonological patterns—such as pronunciation, intonation, or rhythm—towards one another. A key phonological distinction between English and Spanish arises primarily in articulating word-initial voiceless stops, namely /p/, /t/, and /k/. Positive VOT, or aspiration, is characterized by a puff of air following the release of the stop.

A phonological rule of English specifies that word-initial voiceless stops are aspirated or have a positive VOT. Word-initial voiceless stops in Spanish are typically not aspirated or aspirated, much less than English word-initial voiceless stops (Piccinini & Arvaniti, 2015). We will explore Voice Onset Time in code-switching contexts. Code-switching “is a practice common among bilinguals whereby speakers use both languages in a single utterance” (Balukas & Koops, 2015). We will analyze whether English-Spanish bilingual college students demonstrate phonological convergence of VOT in code-switching sentences.

Methods

We will first have all participants complete the demographic and self-report questionnaire about their linguistic background. The questionnaire will include information about their L1 and L2, how often they code-switch, their proficiency in speaking and reading both languages and if they speak any other languages. Lastly, we will include sociocultural questions about their language use, including how positively or negatively they feel about speaking Spanglish and if they have ever been discouraged from speaking Spanglish.

The participants will read “The Rainbow Passage,” a context that includes a mix of English-only sentences, Spanish-only sentences, and sentences that switch from English to Spanish or Spanish to English. We will account for individual idiosyncrasies while striving to derive principles that are as generalizable as possible. The participants will read the following paragraph adapted from the Rainbow Passage with pre-written code-switched sentences:

When the kind sunlight strikes cold raindrops in the air, they act as a prism and form a rainbow. El arcoiris divide la luz blanca en muchos colores hermosos. Estos toman la forma de un arco tan largo, with its path high above, and its two ends apparently beyond the horizon. There is, according to legend, a boiling pot of gold at one end. People look, pero nadie lo encuentra. Cuando un hombre busca algo más allá de su alcance, sus amigos pueden decir que está buscando la olla de oro al final del arcoíris. Throughout the centuries, people have explained the rainbow in various ways. A recording device will be used to record samples of participants using Praat to measure the phonetic output of the individuals and the length of VOT across all conditions. The English words from the Rainbow Passage that we will analyze include kind, cold, two, to, pot, and people. The Spanish words from the Rainbow Passage that we will be analyzing include colores, toman, tan, pero, cuando, and pueden.

Results and Analysis

Several group members contacted qualified bilingual participants and provided access to the prerequisite questionnaire. After collecting basic information about their first language acquisition (FLA) and second language acquisition (SLA) or second language learning (SLL), we distributed the altered Rainbow Passage to participants, ensuring that the recordings of their utterances were clear and suitable for further analysis. We then reviewed all recordings for quality and completeness before forwarding them to those responsible for data analysis. Audio recordings were collected and analyzed using Praat to find the VOT length. The data collected using Praat for each of the 16 participants is grouped into different categories in order to test the following hypotheses.

1. Average bilingual /p/, /t/, /k/ VOT compared to monolingual /p/, /t/, /k/ VOT (BYU)
2. Group English L1 speakers /p/, /t/, /k/ VOT and compare to Spanish L1 speakers /p/, /t/, /k/ VOT
3. The average /p/, /t/, and /k/ VOT for both languages was based on whether or not they were discouraged from speaking Spanglish.

Based on our data grouping, we discovered that our first hypothesis was false because the English VOT mean values for all 16 of our participants (/p/ 85 ms) (/t/ 75 ms) (/k/ 86 ms) were longer than our baseline measurements. Our Spanish voiceless stop mean values based on 16 participants (/p/ 22 ms) (/t/ 75 ms) (/k/ 27 ms) had a /p/ and /t/ value that was longer than the baseline measures, but the /k/ mean was similar to the baseline data.

Figures 2 and 3—The graph shows the mean VOT in milliseconds for all 16 participants. Figure 2 displays the English VOT values for /p/ (blue), /t/ (red), and /k/ (yellow). Figure 3 shows the Spanish VOT values for /p/ (blue), /t/ (red), and /k/ (yellow). 

We found our second hypothesis partially correct because our Spanish VOT was longer than the baseline data, but our English VOT was also longer than the English baseline. Our English L1 Spanish Voiceless mean VOT values (/p/ 28 ms) (/t/ 22 ms) (/k/ 72 ms) were longer than the VOT baseline data collected from Spanish monolinguals. Our English L1 English mean VOT values were similarly longer than those of English monolingual speakers in the baseline data.

Figures 4 and 5 – The graph shows the mean VOT in milliseconds for all participants with English as their first language. Figure 4 shows the English VOT values for /p/ (blue), /t/ (red), and /k/ (yellow). Figure 5 shows the Spanish VOT values for /p/ (blue), /t/ (red), and /k/ (yellow).

We discovered that our Spanish VOT was longer than our baseline for the /k/ voiceless stop, partially supporting our third hypothesis. Our Spanish L1 English mean VOT (/p/ 51 ms), (/t/ 63 ms), (/k/ 72 ms) was shorter than our baseline, as expected. When we looked at our Spanish L1 Spanish, mean VOT (/p/ 18 ms), (/t/ 18 ms), and (/k/ 27 ms), we discovered that the /p/ and /t/ mean VOT values were longer than the baseline, while the /k/ mean VOT value was identical to the baseline value.

Figures 6 and 7 – The graph shows the English VOT values for /p/ (blue), /t/ (red), and /k/ (yellow). Figures 6 and 7 show the mean VOT in milliseconds for all participants whose first language was Spanish. The Spanish VOT values for /p/ (blue), /t/ (red), and /k/ (yellow) are shown in Figure 6.

Since the participants who were discouraged from speaking Spanish had a lower VOT than those who were encouraged to speak Spanish, our fourth hypothesis was false. In comparison to our encouraged Spanish mean VOT, which is (/p/ 24 ms) (/t/ 24 ms) (/k/ 28 ms), our discouraged Spanish mean VOT was (/p/ 18 ms) (/t/ 18 ms) (/k/ 28 ms).

Figures 8 and 9—The graph shows the average VOT in milliseconds for each of the 16 participants according to whether they had been encouraged to speak Spanglish. The mean VOT in ms for the /p/, /t/, and /k/ values for the participants who received encouragement is displayed in Figure 8. The mean VOT in ms for the discouraged participants’/p/, /t/, and /k/ values are displayed in Figure 9.

Figures 10 and 11 – Both are screenshots of the Praat spectrograms of one of our participants. The VOT of the participant who produced the [t] in the word “toman” is displayed in red in Figure 10 (left). The VOT of the participant who produced the [tʰ] in the word “to” is displayed in red in Figure 11 (right). Speaking Spanglish has been discouraged for this individual, whose first language is Spanish.

Figures 12 and 13 –  Both are screenshots of the Praat spectrograms of one of our participants. The VOT of the participant who produced the [p] in the word “pero” is displayed in red in Figure 12 (left). The VOT of the participant who produced the [pʰ] in the word “people” is displayed in red in Figure 13 (right). Speaking Spanglish has been discouraged for this individual, whose first language is Spanish.

Figures 14 and 15 –  Both are screenshots of the Praat spectrograms of one of our participants. The VOT of the person who produced the [k] in the word “colores” is displayed in red in Figure 14 (left). The VOT of the person who produced the [kʰ] in the word “kind” is displayed in red in Figure 15 (right). Speaking Spanglish has been discouraged for this individual, whose first language is Spanish.

Discussion and Conclusion

This research aims to contribute to understanding phonetic changes during code-switching, focusing on the syllables /p/, /t/, and /k/. The findings indicate that, in addition to phonological changes influenced by language-specific properties, hyper-articulation during the experimental process can also impact the results. Notably, the VOTs of both English and Spanish were lengthened, albeit for different reasons. While previous studies suggested that English VOT should remain stable, the observed lengthening may be attributed to hyper-articulation. This phenomenon occurs when participants consciously articulate more clearly to facilitate phonetic analysis, particularly when instructed to produce distinct utterances for experimental recording. The results aligned with some of our hypotheses, though some biases were observed. Further analysis is required to identify academically sound explanations for these findings.

Due to the difference in language properties, the Spanish VOT is lengthened as predicted. Spanish generally exhibits shorter VOT than English. During code-switching between English and Spanish, the VOT of Spanish syllables was observed to be longer than that of monolingual Spanish speakers, suggesting the occurrence of phonological convergence. However, the lengthened English VOT needed to be accounted for in the hypotheses regarding phonological convergence, necessitating further exploration for alternative explanations.

Based on one of our references (Kasia Muldner,2017), English VOT during code-switching is nearly identical to monolingual speakers, which can be attributed to differences in language-specific characteristics. Consequently, English undergoes minimal phonological changes during code-switching. Our hypotheses align with previous studies, suggesting that despite minor variations, English VOT remains nearly identical to that of monolingual English speakers. However, this experiment’s data showed opposite conclusions, forging us to discover more reliable explanations.

Overall, all VOT measures, except for the Spanish /k/, did not reflect our true monolingual English and Spanish speakers. Baseline data demonstrated an increase in VOT data as the place of articulation was placed further back in the vocal cavity. The /p/ VOT was the shortest, next was /t/ VOT, and the longest was the /k/ VOT in both English and Spanish. Our participants produced a similar VOT for all the English stops (around 80 ms) and all Spanish stops (around 23 ms) These findings suggest that bilingual speakers may default to one length of VOT regardless of the place of articulation, whereas monolingual speakers produce a different VOT length for each stop consonant. Despite previous research, this provides evidence that bilinguals differ in VOT production from monolinguals. 

Another interesting finding was hyper-articulation in our sample, which was characterized by more distinct and easily recognizable pronunciations. This phenomenon is hypothesized to have occurred because participants were instructed to produce clear recordings for subsequent analysis using professional phonetic tools (PRAAT), inadvertently creating a more demanding articulatory environment. Moreover, when comparing English in code-switched contexts with that of monolingual speakers, it is important to consider the inherent variations within English itself. Due to its diverse dialects and accents and a long history of linguistic contact, the definition of  ‘standard’ English is ambiguous. This suggests that the participants in our study may have exhibited different English accents from the outset, potentially influencing the results of our experiment.

Our study encountered a few limitations that affected the study’s results. Firstly, it is a vital variation given that our demographic was relatively small (16 people, with 14 college students), which may produce bias. Furthermore, all of them are self-reported bilinguals, even though with a linguistic background questionnaire, the clarification of “bilingual” remained unclear and contained various things. Our linguistic background questionnaire cannot showcase the diversity and dynamic of bilinguals panoramically and thus may ignore some related variations in code-switching. Secondly, though being collected and quantitatively analyzed, the data per se remained unclear in its accuracy in spontaneous code-switching in natural utterances. Taken together, this research contributes to the research looking into the relationship between code-switching and phonological convergence in Spanish-English bilinguals. The observed VOT lengthening during code-switching contexts suggests a dynamic interaction between languages where bilingual transfer does not have a bidirectional relationship. Bilingual speakers may also adjust their articulation based on the contextual demands of the linguistic environment. Future studies could further explore the implications of phonological transfer and its potential effects on language production in bilinguals.

Appendix I: Linguistic background questionnaire[1] [2] 

(1)  Is English your L1 or L2?

(2) Is Spanish your L1 or L2?

(3) Can you read and speak Spanish?

(4) Can you read and speak English?

(5) Do you speak any other language other than English and Spanish?

(if yes, which languages?)

(6) On a scale from 1-10, how often would you say you code-switch between English and Spanish? (Code-switching is the act of alternating between two or more languages or dialects within a conversation or phrase.)

(7) How positively or negatively do you feel about speaking Spanglish?

(8) Have you ever felt discouraged from speaking Spanglish?

• yes
• no

Appendix II: The Rainbow Passage in English and Spanish

When the kind sunlight strikes cold raindrops in the air, they act as a prism and form a rainbow. El arcoiris divide la luz blanca en muchos colores hermosos. Estos toman la forma de un arco tan largo, with its path high above and its two ends apparently beyond the horizon. There is, according to legend, a boiling pot of gold at one end. People look, pero nadie lo encuentra. Cuando un hombre busca algo más allá de su alcance, sus amigos pueden decir que está buscando la olla de oro al final del arcoíris. Throughout the centuries, people have explained the rainbow in various ways.

References:

Balukas, C., & Koops, C. (2015). Spanish-English bilingual voice onset time in spontaneous code-switching. International Journal of Bilingualism, 19(4), 423-443. https://doi.org/10.1177/1367006913516035

Banov, I. K. (2014, December 1). The production of voice onset time in voiceless stops by … BYU ScholarsArchive. https://scholarsarchive.byu.edu/cgi/viewcontent.cgi?article=5339&context=etd

Castañeda Vicente, M. L. (1986). El V.O.T de las oclusivas sordas y sonoras españolas. Estudios de fonética experimental, 2, 91-110.

Kasia Muldner, Leah Hoiting, Leyna Sanger, Lev Blumenfeld and Ida Toivonen.The phonetics of code-switched vowels, Carleton University, Canada https://journals.sagepub.com/doi/10.1177/1367006917709093

Lisker, L., & Abramson, A. S. (1964). A Cross-Language Study of Voicing in Initial Stops: Acoustical Measurements. WORD, 20(3), 384–422. https://doi.org/10.1080/00437956.1964.11659830

Ojeda, Adriana, Ana De Prada Pérez, and Ratree Wayland. Heritage Speakers and their Language Use: A Phonetic Approach to Code-Switching. University of Florida.

Olson, D. J. (2016). The role of code-switching and language context in bilingual phonetic transfer. Journal of the International Phonetic Association, 46(3), 263–285. https://www.jstor.org/stable/26352311

Piccinini, P., & Arvaniti, A. (2015). Voice onset time in Spanish–English spontaneous code-switching. Journal of Phonetics, 52(Sep), 121–137. https://doi.org/10.1016/j.wocn.2015.07.004 Voice onset time. (n.d.). ScienceDirect. Retrieved November 18, 2024, from https://www.sciencedirect.com/topics/biochemistry-genetics-and-molecular-biology/voice-onset-time

[/expander_maker]

Kevin Kim, Shoichiro Kamata, Karin Yamaoka, Raine Torres, Max Fawzi

Japanese is often erroneously considered a “swearless language”, but anyone who has ever been yelled ‘しね’ (meaning ‘to die’) will confidently tell you that like all languages, Japanese has diverse ways of encoding abusive language. In Japanese ‘しね’ only becomes abusive language when in the context of being an insult, but in everyday situations the word simply means ‘to die’ without any connotation of insult. English differs from Japanese by having explicit profanities that carry a vulgar meaning independent of its usage context or syntactic environment. We conducted the following research to discover the discrepancy of semantic typology between Japanese and English profanities or abusive language, and if bilingual speakers endow varying emotional intensity to English profane lexica compared to Japanese abusive language. Our study shows that L1 Japanese L2 English bilinguals view English profanities as less offensive than their L1 English counterparts, report using these English profanities more frequently, and view the equivalent Japanese abusive language as more offensive.

[expander_maker id=”1″ more=”Read more” less=”Read less”]

Introduction and Background

Languages differ significantly in how they encode and express emotions, particularly when it comes to the use of profanity as a means to express anger (for example in the context of an argument). This study examines profanities as a distinct subset of lexicon characterized by inherently vulgar or aggressive meanings, distinguishing them from other words or expressions that acquire such connotations only through context. In other words, in this paper, profanities refer to lexically explicit terms such as ‘fuck’ in English, that convey hostility or vulgarity without the necessity for context, and argue that Japanese lacks such a distinct subset, since words like  ‘しね’ (meaning ‘to die’) can convey aggression or intensity, but lack the explicit vulgarity found in English profanities. This study thus examines and defines profanities as vulgar lexica (found often in English) where its vulgarity is unaffected by its syntactic environment or context; distinguishing them from abusive expressions that only acquire offensive undertones within specific contexts (found often in Japanese). More specifically, Japanese speakers use “various markers of register rather than the explicit deployment of dysphemistic lexical items or expressions” (Jackson & Kennet 2021,1), where connotation of abuse is only endowed through syntactic and environmental inferences (context in which it is being used) even though there can be contextually impolite vocabulary derived from anatomical, excretory, and sexual sources like in English (Hoshino 1971, 31). We will refer to the vulgar and aggressive lexica of English as ‘profanities’ whereas ‘abusive language’ will be used for their Japanese counterpart lexica.

We argue that current studies on Japanese do not accurately reflect empirical use of abusive languages by native speakers. Medias (e.g. anime) that use more performative language tends to distort the perception of how Japanese speakers use language as media lexicons are often unreflective of native speakers’ lexicons. Through this study, we aim to focus on what native speakers consider to be consistently employed in their everyday language.

With our definition of profanities, it assumes a discrepancy of linguistic equivalence between English profanities and Japanese abusive language in its aggressiveness in its use. Therefore, it raises a question; How do Japanese L1 and English L2 speakers juggle the two discrepant lexical categories of ‘profanities’ and ‘abusive language’ and navigate through the lack of equivalence as a bilingual?

We aim to gain an understanding of the cognitive processes that occur for bilingual speakers, when conceptualizing constructs that are not lexically represented in their L1 and discrepancies in linguistic equivalence.

A research by Dewaele (2016) offers a valuable insight into self-reported perception of English swear words by L2 English speakers. The findings show that English LX users (non-native English speakers) over-estimate the offensiveness of most negative emotion-laden words, and avoid using most offensive emotion-laden words. On the other hand, a relevant study by Gawinkowska et al. (2013) found that bilingual speakers are more hesitant to use strong expletives in their L1 than in their L2, likely because they feel less bound by normative influences when speaking in their L2. As such, existing research has presented conflicting results, adding layers of complexity to bilingualism and emotional perception of abusive language in L2 languages.

We expect cultural aspects to play a role in the perception as well; Japanese cultural norms generally discourage direct expressions of intense emotion. One way this manifests is through face, as in outward appearances. Lin & Yamaguchi (2011) state that Japanese culture places greater importance on saving face, as face in collectivist cultures is concerned with an individual’s position in the social hierarchy rather than personal achievement like in individualistic cultures; there is also a tendency to avoid conflict and maintain interpersonal harmony in collectivist cultures. By directly expressing strong emotions, one would lose face in Japanese society and thus there is a social pressure against directly expressing intense emotion. Synthesizing the relevant studies and Japanese cultural aspect, we hypothesize that L1 Japanese speakers will perceive English profanities as being less emotionally charged than their L1 English counterparts. We can also expect L1 Japanese speakers to rate Japanese abusive language roughly the same as their English profanity counterparts or as less offensive overall.

Methods

2.1 Participants

The population of our study is L1 Japanese L2 English speakers, with our control group being L1 English speakers having no knowledge of Japanese. L1 Japanese L2 English bilingual participants are Japanese nationals who did not grow up speaking English at home; their English proficiency were established in the demographic section of the survey including questions regarding their language background. Furthermore, all participants are current college students for ease of recruitment and consistency, putting the age range between 18-28 years old, an age group which likely has the same general vocabulary and therefore slang/profanities, allowing for more homogenous data.

2.3 Design

The survey method was chosen to measure perceptions of both Japanese abusive language and English profanity. It consisted of two phases, where phase 1 focuses on the translation and interpretation of English profanities, and phase 2 focuses on the emotional charge and frequency of use of profanities.

2.3.1 Phase 1

Japanese media often, for performative effect, distorts how abusive language is used; this however does not reflect how native speakers use abusive language. We thus established a list of 10 Japanese abusive expressions with our own English translations, and asked a group of L1 Japanese L2 English speakers (N₁ = 14, 7 male, 7 female) to translate the predetermined English words into Japanese. The results became a basis for the subsequent phase and also a tool to identify abusive lexica that native speakers (18~28) actually use.

2.3.2 Phase 2

A separate survey was sent to both our control group (N₂ = 16, 8 male, 8 female) and to our experimental group (N₃= 13, 3 male, 10 female). The survey asked participants to rate on a 5-point scale (1 = very low, 5 = very high) with regards to a given abusive expression/profanity (1) how well they understand the meaning, (2) how offensive/emotionally charged it is, and (3) how frequently they use it. The control group of L1 English speakers rated solely English profanities, while our experimental group rated both English profanities and Japanese abusive language.

2.4 Manipulation            

To standardize the implied usage context of the profanities (as emotional intensity could be influenced by the varying situations that participants imagine its usage to be), every survey were presented with a GIF of two men arguing, to standardize the perception that the profanities/abusive language in question are used by a person demonstrating anger. Moreover, phase 2 had two versions of the same survey, where Version 1 asked participants to rate English profanities first and of Japanese abusive language second, and vice versa. This was done in hopes of mitigating anchoring effects or priming effects that could bias our findings.

Figure 1: GIF presented to participants during phase 2 survey 

Results and Analysis

3.1 Phase 1 Results

This phase primarily served to establish the abusive language in Japanese that reflected the actual terms used by native speakers. A key finding was in regards to the perception of “shit” and “fuck”: L1 Japanese speakers tend to view “shit” and “fuck” as synonymous when translated into Japanese. Approximately 86% of respondents translated both “shit” and “fuck” as 「クソ」(kuso). This pattern likely reflects the absence of a direct Japanese equivalent for the word “fuck.” Instead, respondents opted for 「クソ」(kuso), which semantically corresponds more closely to “shit” as it denotes fecal matter. This semantic overlap highlights a gap in linguistic equivalence, influencing the way these terms are interpreted and used in the context of L1 Japanese speakers.   

Another key finding was that the translations for “dumbass” and “moron” were predominantly split between two responses: 「バカ」(baka) and 「あほ」(aho). Slightly more respondents translated “dumbass” as 「バカ」(baka), while “moron” was slightly more frequently translated as 「あほ」(aho). For the subsequent phase of the study, we selected the more commonly chosen translation for each term, though the differences were minimal. This suggests that, for Japanese speakers, profanities targeting intellectual levels exhibit limited variability in their translations, possibly reflecting a convergence in semantic interpretation.

The patterns presented were indicative of the nuanced ways in which abusive language is conceptualized and translated by L1 Japanese speakers. Specifically, the findings highlight the challenges of mapping English profanities to Japanese equivalents due to cultural and linguistic differences.

3.2 Phase 2 Results

Quantitative analysis of phase 2 reveals that on average L1 Japanese bilinguals view English profanities as 0.73 less offensive on the 5 point scale than L1 English speakers, with our experimental group rating the 10 English profanities as an average of 2.73, and the control group rating them as 3.45. The largest differences were found in “Ugly Bitch” (difference of 1.76) and the more commonly used “Fuck you” (difference of 1.42), which reflects a significantly different attitude towards even more common swearwords. Our experimental group also viewed the two most commonly used profanities “Shut Up” and “Fuck” as more offensive, with small differentials of 0.47 and 0.33 respectively. Our experimental group however, also rated the selected 9 Japanese translations at an average of 3.70 on the offensiveness scale (i.e. 0.25 higher than the English profanities), with the most offensive being 死ね, equivalent of “Fuck you” with a rating of 4.69 on the offensiveness scale.

Figure 2: Summary of the offensiveness of English profanities and equivalent Japanese abusive language

When it comes to frequency of usage, L1 Japanese bilinguals also reported using the English profanities at a higher frequency, with self-reported frequency being 0.87 points higher than their L1 English counterparts. This was the case for almost all English profanities across the board with the exceptions of “Fuck” and “Ugly Bitch” which had near equal usage; the largest differentials were found in “Moron” (difference of 1.48) and “Dumbass” (difference of 1.35). Interestingly, L1 Japanese speakers reported a rather low frequency of usage for the counterpart Japanese abusive language, with an average of 1.85 on the 5 point frequency scale. The most frequently used one being 「バカ」equivalent to “Dumbass” at 3.08, followed farther behind by 「クソ」 equivalent to “Fuck” at 2.38. This reflects the less frequent use of abusive language in Japanese compared to English which uses profanities more freely. This also corroborates the higher offensiveness of Japanese abusive language, which is used more infrequently and thus packs more of a punch.

Figure 3: Summary of the self-reported frequency of usage of English profanities and equivalent Japanese abusive language

Discussion and Conclusion

4.1 Phase 1 Discussion

In Phase 1, we found that L1 Japanese speakers used the same word 「くそ”」 (kuso) as an approximation for both “shit” and “fuck”. While this was not something that we had initially hypothesized, our findings in Phase 1 supported the idea that the Japanese language has less explicit profanities as previously found in research such as Jackson & Kennet (2021), highlighting the semantic overlap in the translation of “shit” and “fuck”, which underscores the limited availability of distinct Japanese terms that capture the unique connotations of these words in English. Similarly, the minimal variability in the translation of “dumbass” and “moron” suggests that Japanese speakers may perceive insults related to intellectual capability as largely interchangeable, pointing to a possible variability from L1 English speakers in how such terms are employed and understood.

4.2 Phase 2 Discussion

In Phase 2, we found that L1 Japanese speakers viewed English profanities as less offensive than L1 English speakers in all but two instances (“Shut Up” and “Fuck”). This aligned with our hypothesis that L1 Japanese speakers would perceive English profanities as being less emotionally charged. However, this contrasted with the findings of the Dewaele (2016) study which found that non-native English speakers would overestimate the offensiveness of English profanities and avoid using them. Our findings, instead, seemed to align more with Gawinkowska et. al (2013) which found that bilingual speakers were actually more inclined to use strong expletives in their L2 because they felt less bound by the norms of their L1. In our study, we saw evidence of this from the higher rates of usage for English profanities by L2 Japanese speakers than L1 English speakers. Japanese L1 speakers also rated the Japanese approximations of English swear words as being higher in offensiveness than the English swear words themselves. In the midst of conflicting research, we found that our findings followed similar patterns to Gawinkowska et. al. This may be due to the deeply ingrained collectivist cultural norms and emphasis on face-saving that exists in Japanese culture (Lin & Yamaguchi 2011) that may make L1 Japanese speakers more sensitive to abusive language in their native tongue.

4.3 Overall Discussion

This study offers some insight into the perceptions of emotional words in an L2. In our case, this refers to the use of profanities that are explicit without context for speakers who are familiar with abusive language that require context to be abusive. Research that helps understand these experiences will hopefully contribute to a wider understanding of interpersonal connection by identifying the ways language can cause miscommunications due to differences in culture as well as the language itself. Interestingly, additional observations suggest that non-English speakers often react with surprise or discomfort when exposed to the meaning of English profanities. In a street interview conducted by Asian Boss (LINK), Japanese people shared that they find such language uncomfortable and showed that abusive words are rarely used in Japanese culture. This observation also occurs among English speakers with Japanese people. According to the discussion in the Podcast “Hapa 英会話 Podcast – 第305回「汚い言葉遣い」”(LINK) , some English speakers try not to use the profanity with Japanese people even when they are talking each other in English. This similarity shows that hesitation from using profanity comes from the differences across cultural backgrounds regarding when it is appropriate to use profane language, which gives insight to the experiences of Second Language Learners as they navigate unfamiliar territories. These reactions further emphasize the cultural and linguistic differences in how profanities are perceived and used between English and Japanese speakers.

While our research was limited to only include college-aged students, future research would ideally expand the scope to include a variety of ages and educational backgrounds. Future research could also compare our findings with Japanese L1 speakers to another L1 language with contrasting conventions of abusive language.

References

Dewaele, J.-M. (2016). Thirty shades of offensiveness: L1 and LX English users’ understanding, perception and self-reported use of negative emotion-laden words. Journal of Pragmatics, 94, 112–127. https://doi.org/10.1016/j.pragma.2016.01.009 

Gawinkowska, M., Paradowski, M. B., & Bilewicz, M. (2013). Second language as an exemptor from sociocultural norms: Emotion related language choice revisited. PLOS ONE, 8(12), e81225. https://doi.org/10.1371/journal.pone.0081225

Hoshino,  Akira  (1971). Akutai  mokutai  kō  –  akutai  no  shosō  to  kinō. [Thoughts  on  Abusive  Language.  Aspects of Abusive Language and its Functions]. Kikan jinruigaku  2(3): 29– 52.

Jackson, L., & Kennett, B. (2021). Slang and taboo language: An analysis of swearing guides for L2 Japanese learners. Electronic Journal of Contemporary Japanese Studies, 21(2). Retrieved from https://japanesestudies.org.uk/ejcjs/vol21/iss2/jackson_kennett.html

Lin, C.-C., & Yamaguchi, S. (2011). Effects of face experience on emotions and self-esteem in Japanese culture. European Journal of Social Psychology, 41(4), 446–455. https://doi.org/10.1002/ejsp.817

Relevant Info: Hapa 英会話 Podcast. (2020).  第305回「汚い言葉遣い」[Podcast]. Retrieved from https://podcasts.apple.com/jp/podcast/hapa%E8%8B%B1%E4%BC%9A%E8%A9%B1-podcast/id814040014?i=1000491739815  Asian Boss. (2018). Japanese React to English Swear Words [YouTube video]. Retrieved from https://www.youtube.com/watch?v=KHpw5p8qCrI&t=61s

[/expander_maker]