Context-Dependent Script Choice in Emails: The Case of Sumimasen

Introduction

Contemporary written Japanese language mainly uses four scripts: kanji, hiragana, katakana, and the Roman alphabet. For each of these scripts, there are broad, socially accepted conventions for the kinds of words they can be used to express. For example, kanji is said to be used for content words of Japanese and Chinese origin, hiragana for grammatical particles and inflection, and katakana for loanwords from languages other than Chinese. The Roman alphabet has a supplementary role to represent items such as acronyms, foreign words, and units of measurement. These are the conventions taught in introductory Japanese textbooks such as Genki (Banno, Ohno, Sakane & Shinagawa, Citation2004: 18), and demonstrated in a collection of government documents published by the Agency for Cultural Affairs.Footnote¹ However, in real-life situations, usage outside of these conventions occurs in all four script types (Konno, Citation2013). Katakana in particular is often used for non-loanwords, and applied to words of all types and origins in both informal and formal contexts. For example, the Chinese-origin word kantan (easy), or the Japanese-origin word gomi (rubbish), which are typically expressed in kanji and hiragana respectively (as 簡単 and ごみ), can sometimes be seen in katakana (as カンタン and ゴミ) (Masuji, Citation2015b).

Numerous studies have attempted to explain why such variant use of the katakana script may occur.Footnote² Many of these explanations are based on practical considerations, such as placing emphasis on a particular word (in a hiragana or kanji-heavy sentence) or convenience (the katakana representation of a word usually has fewer strokes than the kanji representation) (see Kataoka, Citation1997: 115; Cochrane, Citation2019: 227–230). However, more socially and contextually based motives have also seen attention in contemporary research, especially when it comes to interpersonal communication or persuasive writing (such as text seen on television programs or advertisements) (Masuji, Citation2013; Masuji, Citation2015a, Citation2015b). Of these, the two motives that are of concern to the current study are the issues of context (for example, the situation or aim of communication, or the relationship between the interlocutors), and the politeness strategies employed by the writer, in response to the context (Masuji, Citation2013). That is, the idea that the motive behind a marked use of katakana and the meaning(s) the script use creates can be affected by both context and politeness-related concerns in interpersonal communication (including letters, emails or other forms of online messages).

The current research is also methodologically different from many existing studies in two main ways. First, it departs from the traditional researcher-focused interpretation method to one focusing on how L1 readers interpret script play. Most research on writing-restricted variation to date has explained marked language use through either the researchers’ personal interpretations or through identifying trends via corpus linguistics methods (Robertson, Citation2021). The data for the current study instead draws upon a questionnaire survey targeted at 210 Japanese-L1 undergraduate university students. Secondly, the study focuses on the variant representation of a single word rather than the variant use of a script in general. All of the survey reading passages were identical except for their representations of the word sumimasen (‘sorry’ or ‘excuse me’), a common expression in both written and spoken Japanese. While Robertson (Citation2017) and Kunert (Citation2020) have discussed word-restricted variation with regards to pronouns and loanwords respectively, how word choice intersects with script choice remains an under-studied question. In attending to these two novel angles, this article contributes to the topic of how script use and vocabulary choice interact to create meaning.

From here, I will first describe the background of contemporary katakana use and the term sumimasen. I will also briefly introduce some theoretical concepts relating to politeness, including those from Brown and Levinson’s (Citation1987[1978]) seminal work, which I will be using to interpret the data. Then, I will describe the research methodology and present the data and my analysis, showing how script choice can be part of a wider politeness strategy in Japanese emails.

Katakana use in contemporary Japan

As mentioned in the introduction, katakana is one of the primary Japanese scripts. Like hiragana, it is a phonetic script made up of 48 distinct characters that were originally derived from kanji (Seeley, Citation2000). As the national guidelines do not go into much detail when it comes to katakana use, mass media organisations such as newspaper agencies and television broadcasters have their own published standards for katakana use (for example, NHK kanji hyōki jiten, Citation2011; Kyōdō, Citation2016). According to these standards, katakana is mainly said to have the role of expressing words in the following eight categories (Masuji, Citation2018: 2–3):Footnote³

1. Loanwords and foreign words, including names of people and places, for example, オーストラリア (ōsutoraria, Australia)

2. Onomatopoeia and mimetic expressions, for example, ウロウロ (urouro, aimless wandering)

3. Names of animals and plants, especially in scientific contexts, for example, ネコ (neko, cat), サクラ (sakura, cherry blossom)

4. Colloquial and slang expressions, for example, インチキ (inchiki, fraud)

5. Jargon, including scientific terminology, for example, ケイ素 (keiso, silicon), ヲコト点 (wokoto ten, diacritic marks used in classical kanbun texts to indicate particles and inflections)

6. Some ‘counters’, or josūshi, for example, 3 ケタ (san keta, three digits), or the ko in 2コ (niko, two pieces)

7. Symbols and shapes, for example, the ko in コの字型 (ko no ji gata, ‘U-shaped’), bullet points (アイウ or イロハ instead of a., b., c., etc.)

8. Japanese-language proper nouns, for example, トヨタ (Toyota), ヒロシマ (Hiroshima)

Of the eight categories above, loanwords and foreign words remain the most observable use of katakana in contemporary Japan. In a recent study of magazines, newspapers and television advertisements, Igarashi (Citation2012: 18) found that 91.09% of the extracted katakana-represented words were loanwords. In the case of academic journals, which are considered more formal texts, 98.1% of all katakana-represented words were found to be loanwords or foreign words (Masuji, Citation2018: 8). Most prior studies on ‘non-standard katakana use’ have defined ‘standard katakana use’ as katakana representations of any word that is a loanword, onomatopoeia, or an animal/plant name, and focused on investigating the use of katakana for other word categories. Masuji (Citation2019: 144–145) lists 47 such prior studies in the Japanese language, which have identified many motivations for such non-standard katakana usage, such as focusing on the sound that is being expressed (including pronunciation variants) (Mabuchi, Citation2017) or expressing a difference in nuance to the standard meaning of a word, as written in kanji (Kiko, Citation2007).

However, most Japanese L1 speakers do not refer to such published standards or studies when they use katakana. In reality, Japanese L1 speakers’ understanding of how to use katakana, or even when katakana can be used, is mainly based on two experiences: 1. their compulsory Japanese-language education at school, and 2. how they see words written in the mass media on a daily basis (which is based on the published standards mentioned above). It could be said that a kind of ‘katakana sense’ for standard written Japanese is therefore developed through exposure to such representations from a young age, which gives readers cues as to which representations are common and acceptable (this is similar to the process of ‘language socialization’ described in works such as Burdelski and Cook (Citation2012) and Cook (Citation2012), where sociocultural and linguistic knowledge such as the politeness and appropriateness of certain utterances is acquired through repeated participation in language-based social interactions). It should be noted here that this ‘katakana sense’ can thus be expected to be fairly uniform across Japan (unlike dialects in spoken Japanese, for example), due to the standardised nature of both the compulsory education curriculum and the guides used by mass media organisations.Footnote⁴

Sumimasen in contemporary Japanese

The everyday expression chosen for this article, sumimasen, has multiple meanings in contemporary Japanese. Based on real-life conversation data, Kimura (Citation1994) found five different functions: 1. ‘request marker’, 2. ‘attention-getter’, 3. ‘closing marker’, 4. ‘regret marker’, and 5. ‘gratitude marker’. Similarly, the Digital Daijisen, a popular and frequently updated Japanese language dictionary, defines sumimasen as follows:

As can be seen in the examples above, the expression sumimasen can be used to express both apology and gratitude, which are two very different linguistic acts (see Obana, Citation2015 for detailed exploration of the differences between sumimasen and arigatō as expressions of gratitude). Due to this wide range of potential uses, sumimasen is employed at an extremely high frequency, surpassing other common ways of expressing apology like gomennasai and mōshi wake arimasen. Satō (Citation2011), for example, extracted apologetic expressions from scripts for television and radio serials produced between 1999 and 2009, and found that sumimasen was the most frequently used out of these expressions. Additionally, sumimasen is considered less formal than the other two expressions (Masuji, Citation2013). This ability to express casual apology and gratitude is likely to be the reason why the expression is used so often, and across many different situations.

However, despite its frequency of use, the representation of sumimasen is far from settled. And while much discussion of the meanings and uses of sumimasen exists, this latter concern has been completely overlooked in prior research. For instance, in the Digital Daijisen, the term is written in kanji, as 済みません. However, this kanji-represented form is rarely seen in everyday life, because everyday expressions tend to be written in hiragana.Footnote⁵ For example, gomennasai also has a kanji representation (御免なさい), as do greetings such as konnichiwa and ohayō (今日は and お早う, respectively), but these representations are all uncommon. On the other hand, although these everyday expressions do not fit into the eight categories of katakana use outlined in the previous section above, they are sometimes written in katakana, especially in casual contexts. For example, 39 out of 210 undergraduate students I surveyed said that they use katakana for expressions such as gomen and yoroshiku (another common expression for gratitude) either ‘often’ or ‘sometimes’ in online communication (see also Robertson, this issue). The actual figure for katakana usage is likely to be even higher, as this figure only represents the people who recall using katakana consciously.

The characteristics of frequency and possibility for script choice make sumimasen an ideal object of study for how contextual factors may influence Japanese readers’ understanding of word-targeted script choice. Furthermore, apologies are acts of which authenticity hinges on ‘a wide repertoire of cues to signify remorse’ (Hornsey, Wohl, Harris, Okimoto, Thai, & Wenzel, Citation2020: 367). Apologies also operate in a manner distinct from many other communicative situations, making their representation likely to stand out and be a subject of metalinguistic attention in Japan. Using terminology from Brown and Levinson’s politeness theory, the apologetic usage of sumimasen is a ‘negative politeness strategy’ by the speaker or writer, as it seeks to mitigate the threat to the addressee’s ‘face’. Brown and Levinson postulated that ‘all competent adult members of a society’ cooperate with each other to defend their ‘face’ (the ‘public self-image that every member [of a society] wants to claim for himself’), from ‘face-threatening acts’, (FTAs) (1987: 61). Language users adjust their politeness strategies depending on the seriousness, or ‘weightiness’ of the FTA. This weightiness is calculated as the sum of the social distance (D) between the speaker and the addressee, the power (P) the addressee has over the speaker, and the degree to which the FTA is rated an imposition in that culture (R). As sumimasen is a form of casual apology, as explained above, it is likely to be used when the weight of the FTA is considered relatively low (compared to gomennasai and mōshi wake arimasen).

The politeness strategy that is used to defend face can be either negative or positive. Simply put, ‘positive politeness’ is about respecting a person’s wants to ‘be ratified, understood, approved of, liked or admired’ by others, while ‘negative politeness’ is about respecting a person’s wants ‘that his actions be unimpeded by others’ (1987: 62). Compared to negative politeness strategies, positive politeness strategies are said to be used in situations that involve lower-level FTAs (1987: 60). However, this categorisation of positive versus negative politeness has been criticised by many scholars over the years (Cook, Citation2011; Ide, Citation1987, Citation1989; Mimaki, Citation2002). Ide (Citation1987), for example, famously pointed out that much of polite language use in Japanese (especially honorifics) is done according to social conventions and motivated by ‘discernment’, rather than being a result of such considerations to reduce the impact of FTAs. Due to this, strategies that are considered part of ‘negative politeness’ by Brown and Levinson (such as the use of honorifics) can often coexist with ‘positive politeness strategies’ in the Japanese language. Taking into account these criticisms, in this article, politeness was seen as being created by two types of language use, as outlined in Mimaki (Citation2002): strategic language use (for example, making jokes to make the other party comfortable), and language use based on social norms (for example, one must use keigo when meeting someone for the first time). Although Mimaki’s terminology is more applicable to the Japanese language, I will also refer to the original terms by Brown and Levinson when I am discussing matters that are relevant to the wider global research on politeness theory.

Although Brown and Levinson’s focus is on spoken language, my hypothesis is that Japanese writers also find methods restricted to the written mode for adopting politeness strategies in apologies to mitigate the impact to the reader’s ‘face’, and that they adjust these strategies according to the seriousness of the FTA. For example, Masuji (Citation2013) suggested that the exchange of katakana-represented sumimasen in work emails may be an example of a ‘strategic language use’ where nonstandard language was used to create a sense of camaraderie, and that it was more likely to be employed when the social distance (D) between the sender and the recipient was close, and the seriousness of the FTA was thus relatively low. In other words, despite apologies generally being considered as a negative politeness strategy (‘Strategy 5’ in Brown & Levinson, Citation1987[1978]: 187–190), the katakana representation of sumimasen may be a positive politeness strategy on the word level that works by ‘minimizing the size of the face threat’ and emphasizing the mutual shared interest between the interlocutors (similar to Brown and Levinson’s ‘Strategy 11: Be optimistic’: 126). However, as this hypothesis in the 2013 study was only based on two test subjects and their work emails, this article will investigate the matter further with more test subjects and in differing contexts.

Method

The aim of the current article was thus to investigate how difference in contextual factors influence the perception of katakana- and hiragana-represented sumimasen by L1 speakers. The research was conducted over a period of three months, between July and October 2018. The survey respondents were 210 Japanese-L1 speakers, who were all undergraduate (age 18–21) humanities majors at Aichi Shukutoku University. The survey was distributed on paper, some distributed in a classroom setting and the rest handed to students individually.Footnote⁶

All survey respondents were asked to imagine themselves being part of a group of university students who are working towards a common goal. Four distinct contexts within this team setting were given to all survey respondents, with each context describing an interaction with a person the respondent had recently met. In two of the contexts, the respondent was told they had asked the person to complete a very time-consuming data aggregation task. The person realised 12 hours before the deadline that they probably would not finish the work by the deadline, however, and emailed the respondent to apologise. In one of these two settings the person writing the email was described as younger than the respondent, and in the other they were the same age. In the other two contexts, the situation was the same, but the relationship was reversed. The respondent was asked to imagine that someone else in the team had asked them to do the data aggregation task. However, the respondent themselves realised 12 hours before the deadline that they would not make the deadline, so they wrote an email to apologise. Again, this setting was divided into two contexts depending on whether the person they were asked to write to was older, or of the same age. In all four contexts, survey respondents were asked to imagine that both parties involved were of the same gender, to minimise the influence of gender-related concerns in how individuals interpreted representations of sumimasen. While the question of how the gender of an interlocuter influences the use and understanding of script practices is an interesting area for future research, and gender-based divides in script practices have been noted throughout Japan’s history (see Yoda, Citation2004; Kataoka, Citation1995; or Robertson, this issue), considering gender here would limit the ability of the analysis to focus on the politeness-related concerns key to our discussion. For clarification purposes, the four contexts are shown in Table 1 below.

Survey respondents were then shown two emails for each context. Both emails read: ‘I don’t think I’ll get the data in on time. I’m sorry (sumimasen). [Name of sender]’. However, one email had sumimasen in hiragana (Email A), and the other had sumimasen in katakana (Email B). For each of the four contexts, there were 10 questions in which respondents were asked to compare the emails around descriptors such as politeness, friendliness, and neutrality.Footnote⁷ For each of these questions, respondents were asked to pick Email A or B on a 5-point Likert scale (A, more A than B, neutral/I don’t know, more B than A, B). The ten questions are reproduced below (the Japanese wording for questions 1 and 9 differ depending on the context, and there are two options for question 10):

Q1. Which email feels more apologetic (shazai no ito)?

Q2. Which email feels more thought-out (teinei)?

Q3. Which email feels more insincere (fuseijitsu)?

Q4. Which email feels more original (shinsensa)?

Q5. Which email feels more friendly (shitashimi)?

Q6. Which email feels more neutral (bunan)?

Q7. Which email feels more polite (reigi tadashii)?

Q8. Which email do you dislike?

Q9. Which email is used more often in this kind of relationship?

Q10a. Which email makes the sender more likeable to you (kōkan wo motsu)? [for C1 and C2]

Q10b. Which email would you send if you want to build a good relationship with the recipient? [for C3 and C4]

To summarize the objectives of the research using Brown and Levinson’s FTA formula, there are two set levels of ‘weightiness’ depending on whether the relationship between the sender and the recipient of the email is hierarchical or not. Of the three variables in the FTA formula, the social distance between the two parties (D) and the rating of imposition (R) are relatively ‘fixed’, although individual perceptions of such factors are inevitably subjective: the email exchange takes place between two people who have just met (a relationship where polite behaviour is usually observed), and last-minute report of the failure to complete a complicated task would be perceived by most as a high level of imposition for both parties. The main difference between C1/C3 and C2/C4 is the ‘power’ relationship (P) between the recipient and sender of the email. This allows for an investigation of how the P variable (and thus the overall weightiness of the email apology) influences script choice by the respondents.Footnote⁸ Furthermore, comparison of respondents’ individual answers allowed the survey to also investigate whether or not survey respondents would have the same preferences when sending an email, as when they receive the email themselves.

Findings

The results were quantified using the 5-point Likert scale described above. ‘A’ was rated as one point, ‘Maybe A’ was rated as two, ‘Neutral/I don’t know’ as three, ‘Maybe B’ as four, and ‘B’ as five. A mean score of likelihood of katakana preference (M) was then calculated. The higher M is for a particular question, the more respondents chose the email with the katakana sumimasen. For example, in Table 2, the M for Q8 (‘Which email do you dislike?’) in Context 1 is 4.17, meaning that respondents were highly likely to choose the katakana sumimasen as their answer in Context 1. The results tables for the other three contexts are in the Appendix.

In this section, I will describe these findings from four main perspectives. Firstly, I will examine how the hierarchical relationship between the recipient and sender of the message may influence M by comparing the two contexts in which the respondent is the recipient of the apology (C1 and C2, Comparison A), then the two contexts in which the respondent is the sender of the apology (C3 and C4, Comparison B). Then, I will examine whether or not the respondent’s role (being the sender or the recipient of the apology) influences M, by comparing the two contexts that involve hierarchical relationships (C1 and C3, Comparison C) and then the two contexts that involve equal relationships (C2 and C4, Comparison D). Paired-samples t-testing was used to determine whether or not the differences in M between the context pairs were statistically significant. As no statistically significant differences were found in any of the context pairs for Question 2 and 7 (on whether or not the email felt thought-out and polite, respectively),Footnote⁹ the analysis below focuses mainly on the eight other questions.

Discussion and conclusion

The aim of the current research was to investigate how contextual factors influence the perception of the hiragana versus katakana sumimasen. It should be noted first that there was a general preference across all four contexts for the hiragana sumimasen, which was generally considered to be more thought-out, neutral, and apologetic. One of the reasons for this preference could be the fact that the research focused on an email exchange between people who have just met, regarding a delay in meeting the deadline for a highly time-consuming task (that is, with a high level of D and R in the FTA formula). Many respondents seem to have concluded that the nature of the apology was thus ‘weighty’, or highly face-threatening, leading them to prefer the social-norm-based representation. This preference for the hiragana sumimasen was generally stronger in situations where the respondent was on the sending end of the apology, rather than the receiving end.

Looking at general trends for the katakana sumimasen across all contexts, some respondents associated the katakana representation with positive characteristics such as friendliness and originality, but others strongly disliked it and perceived it as insincere.Footnote¹⁰ This means that the use of katakana for an apology is accompanied by a high level of interpersonal risk. However, it was preferred by some individuals as part of their strategic language use, to take advantage of the aforementioned associations such as friendliness and originality. This even occurred in the more ‘serious’ FTA context of sending an apology in a hierarchical relationship. In particular, respondents generally felt that the risk of using the katakana sumimasen was lower between people of the same age, compared to a relationship in which there was an age difference. This is consistent with Brown and Levinson’s formula, which shows that an apology would be less face-threatening when the power difference (P) between the two parties is small. Overall, 96 out of 210 participants had responses that differed depending on the P level.

In other words, by writing sumimasen in katakana, one can turn a usually negative politeness strategy (apology) into a positive politeness strategy, for use between same-age students. Brown and Levinson postulated that contractions such as ‘wanna … ?’ instead of ‘do you want to … ?’ in the English language are used for a similar effect; while the standard forms of these expressions are negative politeness strategies that reduce the imposition of the FTA, one can turn them into positive politeness markers that signal ‘in-group membership’ by using their contracted forms (1987: 270–271). This effect has also been observed in the Japanese language with gomennasai (sorry) and gomen (an abbreviated form of gomennasai) (Hidaka, Citation2017). The present article is the first to reveal that script choice can also convert a negative politeness strategy into a positive one in written Japanese, or that script choice can be part of one’s ‘strategic language use’ for politeness considerations, using Mimaki’s terminology. Ogawa (Citation1993: 42) states sumimasen is used for different meanings depending on the relationship between the two parties, the seriousness of the matter at hand, and any age gaps. Script choice is one of the strategies that can be employed to express such differences in nuance.

Another general trend was the difference in answers between contexts where the respondent was sending the email, and the contexts where the respondent was receiving the email. The lack of correlation in the answers for Question 10 showed that, even when respondents found a certain representation to be more likeable in an email that they received, they did not immediately opt to choose the same representation when sending the same email themselves. It is possible to interpret this trend as resulting from a cultural tendency to wish to be perceived as being extra careful and polite towards others – particularly those who are more senior. Usami (Citation2001: 25) criticised Brown and Levinson’s theory for focusing more on the speaker than the addressee, and observed that the speaker and the addressee may reach significantly different calculations of the seriousness of the same FTA. It seems that Japanese university students have a heightened awareness of this potential gap and tend to add a significant buffer to the calculation when estimating how serious an FTA may be perceived by a more senior addressee. Ihasz concluded in his questionnaire research on katakana-represented words that ‘the Japanese are reluctant to use katakana towards those who are more senior, even if the relationship is good and close’ (Citation2010: 27). Ihasz’s research targeted katakana-represented words in general (including loanwords), but his observation has parallels with the results of the present research.

Despite these general trends, however, there was a large level of individual variation in the survey results. We can observe this in a couple of ways. By looking at the mean for Q5 (friendliness), for example, it appears as though respondents were only slightly more likely to choose the katakana sumimasen as being more friendly (the M was between 3.25 and 3.33 for all four contexts). However, Q5 also had a large standard deviation (SD) across all four contexts, showing that there was significant variability between individual responses. Another example is Q9 in comparison C, which showed results that were inconsistent with the aforementioned general trends (respondents believing that the katakana representation was more common when sending an email to an older student, compared to receiving one from a younger student). There are several possible reasons for these inconsistencies and variations. Firstly, different individuals may have different ways of calculating the seriousness of the hypothetical FTA used in the study. Even though D (social distance) and R (rating of cultural imposition) were ‘fixed’ by providing clear instructions in the survey, the individual perception of all three FTA variables is derived through a subjective thought process, influenced by prior experience or other factors. Secondly, survey respondents would have a disposition towards either following social norms or using scripts strategically in their everyday language use, which may be reflected in their survey responses. And lastly, the associations an individual has with the hiragana versus katakana sumimasen are also subjective, which, combined with the above differences in politeness strategies, produces a myriad of individual script choice strategies for sumimasen.

To conclude, this article showed that contextual circumstances such as the relationship between the sender and recipient and the role that one plays in the communication have a statistically significant impact on the perception of script choice for the word sumimasen in Japanese emails. The level of preference for the katakana sumimasen was highest when receiving the apologetic email from a student of the same age, and lowest when sending the same email to an older student. These results are likely due to the respondents’ expectations of politeness strategies being adjusted depending on the context. However, there was a significant number of respondents whose preferences could not be explained by the variables in this investigation. While the study assumed that the ‘katakana sense’ for standard Japanese would be fairly uniform across Japan due to standardised script rules in education and mass media, there may be other factors such as gender and social media use that could shed light on these variations. It is also possible that actual usage of the katakana sumimasen differs from the picture painted by these survey results, where respondents were asked to focus solely on the issue of script choice. To understand why some respondents preferred the ‘strategic use’ of the non-standard and therefore highly ‘risky’ katakana, the phenomenon needs to be investigated further in other situations and social relationships where katakana use may be more common.