Semiotic Analysis of Kokuji Formation

Han characters (kanji) used in written Japanese come from China, but there are characters created separately in Japan called kokuji. This research aims to investigate the formation of kokuji according to Peirce’s semiotics and rikusho (traditional classification of Chinese characters) by Xu Shen

Despite the fact that kokuji can be loosely defined as "kanji which were produced in Japan," there is no established list of standards that a character must fulfil in order to be classified as a kokuji (Buck, 1969). For the sake of clarity, this paper shall cite Seeley (1991), Sugimoto & Iwabuchi (1998), and Nomura & Koike (1992) and use the common criteria that a kokuji must be created in Japan, not China, and that a kokuji must adhere to the principles of Han character formation.
To properly specify the principles of Han character formation, this research refers to 六 書 rikusho, a categorisation of Chinese characters into six classes that explains their creation (Matsumura, 2006). The theory is often attributed to Xu Shen (c. 58-148 CE), a scholar from the Eastern Han Dynasty (Brown, 2014;Lewis, 1999). Rikusho is commonly used as a guide for analyses of Han character structure (Qiu, 2000) and as such was adopted by Japanese scholars such as Tsutomu Sugimoto and Tadasu Iwabuchi to analyse Japanese kanji. Sugimoto and Iwabuchi's Nihongogaku Jiten (1998) lists the six principles as 象形 shoukei (pictographs), 指事 shiji 1 (symbols), 会意 kaii (ideographs), 形声 keisei (semasio-phonetic compounds), 仮借 kasha (phonetic loans), dan 転注 tenchuu (semantic and phonetic loans). Sugimoto and Iwabuchi further defined each of the classes as such: shoukei characters are kanji that imitate the physical form of objects they describe; shiji is kanji that illustrate abstract concepts visually; kaii are formed by combining two or more kanji to represent a new concept; keisei consist of two elements: one phonetic and one semantic element; kasha characters borrow the phonetic value of a kanji to represent a new abstract idea; tenchuu characters change the meaning of a kanji through association and other factors, then utilise it with the new meaning. In addition to the six classes found in the traditional version of rikusho, we also proposed a new category by combining two of the classes: the kaii-keisei class. Examples of the classes are listed in table 1.
To understand the reasoning behind our proposition of a new class, we should first acknowledge that characters in the keisei category are composed of two elements: one phonetic and one semantic (Sugimoto & Iwabuchi, 1998; hence the translation "semasio-phonetic compound" (Henshall, 1998)). In regards to the phonetic element, Henshall (1998) wrote that "[t]he element used as a phonetic was usually chosen from a range of similarly pronounced characters on the basis of its also lending relevant semantic connotations of its own." This means that the phonetic element is not only used for its sound, but also for its meaning, to some extent. Henshall further noted that "[t]hus it is also possible to consider such semasio-phonetics as ideographs" and that the classification of characters into one category or the other fully depends on "whether the semantic role is felt to outweigh the phonetic or vice-versa," making such classifications stand on a rather subjective ground. The fine distinction between kaii and keisei classes can further increase in complexity when a kanji thought to be a kaii character contains an element that is also used for their phonetic value: an element which "could therefore be said to be a phonetic element expressing its own meaning through its own sound" (Henshall, 1998).

Table 1 Kanji categories and their examples
This problem of classification, even for Henshall, is thought to be superfluous since it often does not go beyond mere speculation (Henshall, 1998). In this paper, we strove to avoid categorising such kokuji subjectively (and potentially erroneously) into one category or the other by conducting biased or uninformed evaluation of the characters' phonetic elements. To resolve this issue, we then proposed a new category, the kaii-keisei class, for characters whose elements share both kaii and keisei principles, namely having one or more elements that contributed both phonetic and semantic values simultaneously to the whole character.
From a semiotic standpoint, this research applies the theory of basic sign structure to the formation of signs, (in this case, kokuji). The theory was proposed by C. S. Peirce (1839Peirce ( -1914, an American philosopher and logician (Merriam-Webster, 1960). Peirce defined a sign to be composed of three related parts: an "object", a "sign-vehicle", and an "interpretant" (Atkin, 2013;Peirce, 1998). The semiotic object is an object or idea that the sign is meant to 日 'sun' Imitation of the physical appearance of the sun (Sugimoto & Iwabuchi, 1998). 木 'tree' A picture of a tree (Henshall, 1998). 目 'eye' A picture of an eye (Henshall, 1998).
Over time, the original word for 'ax' and the actual character for 'I' fell into disuse, resulting in the usage of 〈我〉only to mean 'I' (Qiu, 2000). 転注 tenchuu (semantic and phonetic loans) Read as gaku to represent 'music', as in 音楽. Listening to music is a comfortable or fun experience, so the character's reading became raku to mean 'comfort'. Because humans seek comfort, 楽 further changed its reading to gyou and took another meaning: 'to wish; to want' (Sugimoto & Iwabuchi, 1998). symbolise. The "sign-vehicle" differs from the "sign" as a whole in that it's a primary element of the latter that must be present to correctly describe the object (Atkin, 2013). Additionally, it can also contain secondary elements to further describe the object in detail (Atkin, 2013). The sign-vehicle functions by describing the sign's object. Conversely, the object limits the signvehicle by giving it a set of rules on how it should describe the object. The rules are then reflected in the sign-vehicles form. The interpretant, which can be understood as humans' understanding of the sign, is formed by a person's observation of the relation between the two parts mentioned above. It is also affected through the sign-vehicle to form a person's understanding of the object (Atkin, 2013). The relation between the three parts is described in figure 1. This research aims to investigate how and in accordance with what principles are kokuji formed. While numerous studies have been conducted to analyse the formation of Han characters created in China, the amount of research dedicated to kokuji falls significantly behind it. As an illustration, a Google Scholar search of the keyword "kanji" yielded about 217,000 results in the time span of 1 January 2012 and 11 May 2022. Meanwhile, a search of the term "kokuji" from the same time frame returned only about 1220 results, a mere 0.5 percent of the number of articles discussing the former topic. Aside from the potential of giving kokuji more scholarly exposure, this research will also answer whether Han characters created outside of China follow similar creationary fundamentals or deviate from them, given the difference in the creators' linguistic and cultural backgrounds. Semiotic analyses of kokuji can uncover these tendencies and provide an insight into both the Japanese culture and the workings of the Japanese language, as well as testing the validity of character formation theories which are usually only applied to Han characters of Chinese origin.
Past literature pertaining to kokuji within the past ten years include "Kokuji (国字): The Japanese 'National Characters'. A Case Study: The Japanese Acquatic [sic] Fauna" (2015), an article written by Giovanni Borriello, Ph.D., the Professor of History and Institutions of Asia at Roma Tre University, Italy. The article compiled a list of kokuji for names of aquatic animals and divides them based on the zoological categories of the animals they describe. Borriello also noted which of the characters were still in use in modern Japan, their equivalents in Chinese, and if the characters entered the Chinese language, or conversely, if characters of the same meaning coined in China replaced the kokuji in Japan, leading to their obsolescence in modern Japanese. This paper differs from previous research in that it analyses a number of kokuji not limited to a specific category, doing so from a semiotic standpoint, and not comparing the kokuji to Han characters made outside of Japan.
This research aims to do as follows: (1) to describe the methods used by Japanese people to form kokuji; (2) to describe the role of a word's meaning in kokuji formation; (3) to describe what elements can be a kokuji's sign-vehicle; and (4) to describe the reasoning behind the selection of certain elements to be the sign-vehicle. As hypotheses, we believe that (1) with kokuji being a subset of Han characters (Matsumura, 2006;Yamaguchi et al., 2013), kokuji follow the formation principles of Chinese-made Han characters, albeit, with some expected deviations or differences in use owing to linguistic and cultural differences between the language and people of the two nations; (2) referring to Peirce's theory, the meaning of a kokuji has a role in its formation by mediately affecting its sign-vehicle(s) and interpretant; (3) signvehicles of kokuji are components of pre-existing Han characters, in accordance with the fact that they follow Han character formation principles; (4) aforementioned components are chosen due to having sufficient descriptive capabilities and thus passing the limitations imposed by the semiotic object.

RESEARCH METHODS
Library research was used to gather information and data samples regarding kokuji. Samples were taken from the list of kokuji presented in Reiman (1983), itself a compilation of kokuji from Morohashi's Dai Kan-Wa Jiten (1960), Toudou's Kan-Wa Jiten (1982, names of places in Japan, and other sources. Character definitions, translations, and readings were obtained from JMdict/EDICT Japanese-English Electronic Dictionary database (Electronic Dictionary Research and Development Group, 2003), Daijirin, 3 rd edition (Matsumura, 2006), and Oubunsha Kokugo Jiten (Yamaguchi et al., 2013). Kokuji samples, readings, definitions, and their respective rikusho classes are presented in the appendix of this paper.
Data samples were then analysed using two theories: 六書 rikusho, the traditional categorization of Han characters into six categories, and Peirce's semiotic theory of basic sign structure. Characters were first classified into the six categories of rikusho based on the relations between their meanings and components. The kokuji were then examined as semiotic signs to determine the role of their meanings in their formation. Components were also studied to ascertain the nature and the reasoning behind their assignment as the kokuji's sign-vehicles.

RESULTS AND DISCUSSION
There were 87 kokuji data samples taken from Reiman (1983). The 87 samples were distributed in 5 out of 6 rikusho categories: kaii characters made up most of the data and there was no sample that belonged to shoukei class.
Analysis results are presented by rikusho categories, including one additional proposed category not mentioned in Nihongogaku Jiten, with each category represented by one kokuji sample. Samples were analysed based on their formation principles, the role of their meaning in said formation, the components that function as their sign-vehicles, and the reasoning behind their selection as sign-vehicles.

1) Kokuji belonging to shoukei category
Of the 87 data samples, none belonged to shoukei. We believe that the Japanese people did not have any necessity to create kokuji of this type since the need to represent simple objects had already been fulfilled by the use of pre-existing kanji.

2) Kokuji belonging to shiji category
(69) 丼 どんぶり トン、タン、ショウ、セイ donburi ("splash!" sound; rice bowl topped with meat, fish, etc.) 〈丼〉was formed by combining two elements, 〈井〉 'well' and〈丶〉, to symbolize an abstract concept: the sound of a solid object falling into water ("splash!"). This makes 〈丼〉a shiji character. The meaning 'splash!' plays a role in forming the sign by defining its sign-vehicles. 〈井〉is a kanji of an object with a connection to the sound "splash!" and 〈丶〉is a kanji stroke used as a visual element. 〈井〉was chosen to be the sign-vehicle due to three factors, (1) wells are common objects; (2) wells can contain enough water to produce a sound when their contents are disturbed; (3) wells are symbolized with a visually simple kanji, making it more easily combined with other kanji components (as opposed to kanji of other water-filled objects such as 〈桶〉 oke 'bucket', 〈湖〉mizuumi 'lake', or 〈池〉ike 'pool'). 〈丶〉was chosen to represent a solid object falling into the well due to having only a single stroke, in accordance with shiji principle of adding minimal visual elements to a kanji. Moreover, 〈丶〉can represent the concept of an 'object' without imposing overly strict limitations.

3) Kokuji belonging to kaii category
(1) 鱩 はたはた -Arctoscopus japonicus (sailfin sandfish, Japanese sandfish) 〈鱩〉, meaning hatahata or 'Japanese sandfish', is composed of 〈魚〉sakana 'fish' and 〈雷〉kaminari 'lightning'. The Japanese sandfish commonly spawn and are caught in the months with frequent thunderstorms (Nihon Doubutsu Gakkai, 1897). The kokuji for the fish was made through an association between the meanings of two existing kanji, making it belong to the kaii category. As a semiotic object, the fish determines the sign-vehicles for its sign, namely the kanji for 'fish' and 'lightning'. 〈魚〉is the kanji for a group of animals the Japanese sandfish belong to and 〈雷〉is a kanji for a natural phenomenon with a connection to their life cycle. The two sign-vehicles were chosen for their ability to describe the object's traits: their biological category, life cycle, and relationship with Japan's climate.

4) Kokuji belonging to keisei category
(20) 錻 -ブ tin plate 〈錻〉is a kokuji used to spell the word 錻力 buriki 'tin plate'. The semantic element of this kokuji is 〈金〉 'metal' and the phonetic element is 〈武〉, read as bu. 〈錻〉was classified as a keisei character, not kaii-keisei, since the meaning contained in its phonetic element ('weapon; military; soldier') is not used. The word "buriki" is borrowed from Dutch blik (Matsumura, 2006). The sign's object, 'tin plate', defines its sign-vehicles through its meaning and pronunciation. 〈金〉as a sign-vehicle is a kanji element explaining the chemical property of the object. 〈武〉is a kanji with a suitable pronunciation to spelling the word "buriki". The sign-vehicles are used to represent the object's meaning and pronunciation, respectively.

5) Kokuji belonging to kaii-keisei category
Since the kaii-keisei class is a newly-proposed group in this research, the findings related to the kokuji of this class will be discussed in full in this section. To reiterate, the classification criterion is that a kokuji must have the features of both kaii and keisei categories, namely, having one or more elements that contribute both meaning (semantic value) and pronunciation (phonetic value) to the kokuji. There were four kokuji that met this condition.
(42) 裃 かみしも old ceremonial garb; samurai garb Following definitions given in Matsumura (2006) and Bryant & Badgley (2019), kamishimo is a two-piece traditional Japanese outfit, consisting of an upper part (a hitatare or a kataginu) and a lower part, a hakama. The word is written in Japanese with the kokuji 〈裃〉 (Matsumura, 2006). This kokuji follows the kaii principle in that it is formed through semantic association of its elements: 〈衤〉, a component which means 'clothing' (Matsumura, 2006;Yamaguchi et al., 2013); 〈上〉'up' and 〈下〉'down'. The kokuji can also be considered to follow the keisei principle since the elements 〈上〉kami and 〈下〉shimo are used for their phonetic values as well. The clothing as the semiotic object has a role of limiting the signvehicles. The sign-vehicles of this kokuji consist of a character component, 〈衤〉, and two kanji; 〈上〉and 〈下〉. 〈衤〉was chosen as a sign-vehicle for its meaning, which appropriately describes what item a kamishimo is. 〈上〉and 〈下〉became the sign-vehicles due to their ability to illustrate the form of a kamishimo and its pronunciation in Japanese.
This kokuji follows the kaii principle since it describes a new concept using two existing kanji: 〈酉〉 (a Han character component meaning 'sake' or 'alcohol') and〈元〉moto 'source; origin'; while also relying on the keisei principle by using the component 〈元〉as a phonetic element for the character 〈酛〉. The object 'yeast' limits the sign-vehicles to be the kanji components for 'sake' and 'source'. 〈酉〉is a kanji component that tells the use of the object and 〈元〉is a kanji that describes the object's relation to the previous sign-vehicle. The signvehicle 〈酉〉was chosen because yeast is an ingredient used to ferment Japanese rice wine, miso (Japanese fermented soybean paste), and shouyu (Japanese soy sauce) (Machida et al., 2008). 〈元〉was chosen to describe the yeast's purpose in relation to the other sign-vehicle.
(57) 颪 おろし wind blowing down from mountains The 57 th sample from the list was classified as a kaii-keisei character due to the element 〈下〉oroshi 'to lower' that aided the kokuji's formation through both its meaning and pronunciation. The object, a type of wind, limits the form of sign-vehicles present the kokuji's visual appearance. 〈風〉'wind' and〈下〉'to lower' are both pre-existing kanji, with the first being chosen for its meaning and the latter for both its semantic and phonetic values: implication of a downward movement and the reading "oroshi".
The kokuji for 'gland' was categorized into the kaii-keisei owing to 1) its accordance to the kaii principle: being formed through association of the meanings of two elements, 〈⺼〉'meat' and〈泉〉'spring; fountain'; and 2) its conformity to the keisei principle: using 〈泉〉for its pronunciation in addition to its meaning. A gland, the object which is symbolized by the kokuji, serves to define the sign-vehicles. The sign-vehicles of this kokuji are composed of a radical and a kanji. The radical 〈⺼〉was chosen because glands can be described as carnal, visceral organs. Meanwhile, there are two reasons behind〈泉〉's selection as the signvehicle. Firstly, a gland has a semantic similarity with a spring or fountain in that they both let out fluid, be it bodily or geologically. Secondly, it was also selected for the similarity of its on reading with the kokuji's reading: sen. The kokuji〈麿〉maro, an archaic term for 'I; me', is a combination of two kanji, 〈麻〉and〈呂〉. Both of these kanji,〈麻〉(kun'yomi: asa; on'yomi: ma) &〈呂〉(kun'yomi: sebone; on'yomi: ro, ryo), were borrowed for their pronunciations to make a new character. This method of character formation conforms to the kasha principle. The object-the concept of one's own self with an archaic tone-defines the sign-vehicle through its pronunciation, "maro". 〈麻〉and 〈呂〉are kanji with on readings "ma" and "ro" respectively. The two kanji were chosen to be sign-vehicles owing to their suitable readings, enabling them to signify the object's pronunciation in Japanese.

7)
Kokuji belonging to tenchuu category (69) 丼 どんぶり トン、タン、ショウ、セイ donburi ("splash!" sound; rice bowl topped with meat, fish, etc.) 〈丼〉donburi was originally a shiji character signifying the sound "splash!", but the kokuji underwent a shift in meaning to represent 'rice bowl topped with meat' (as in 丼もの donburi-mono). The old meaning, 'splash!', was associated with the sound of rice placed into a bowl ("plop!"), creating a new meaning in the process. This change of meaning complied with the tenchuu principle. Donburi(-mono) as the semiotic object serves to determine the signvehicle through an associated concept (the sound "plop!"). As a sign-vehicle, the character 〈丼〉represents the aforementioned sound by way of visualization: an object falling into a well, making a sound. 〈丼〉was chosen for having a semantic and phonetic association with donburi-mono, the sign's object.

CONCLUSIONS
Results of this study have shown that the Japanese people created kokuji in accordance with the six principles of rikusho. Out of the 87 kokuji samples, no kokuji belongs to shoukei (0%), 2 belong to shiji (2.3%), 62 to kaii (71.3%), 17 to keisei (19.5%), 4 to the newly-proposed kaii-keisei class (4.6%), 2 to kasha (2.3%), and 1 to tenchuu (1%). Regarding the number of kokuji in kaii class, we surmise that the principle of kaii is most easily followed in Japanese. On the other hand, we suppose that none of the kokuji pertain to shoukei owing to the following reasons: (1) shoukei Han characters are essentially pictures resembling the physical form of an object (Sugimoto & Iwabuchi, 1998); (2) the number of objects which can be represented by pictures are limited; (3) kanji originating from China (as opposed to kokuji) have already symbolized many of said objects; (4) shoukei kanji originating from China are readily available and widely used in Japan.
From a semiotic point of view as proposed by Peirce, the meanings of words serve to define and limit the sign-vehicles which shape the form of kokuji. Elements that can function as sign-vehicles are existing kanji or their components. These sign-vehicles represent various things with connections to the object, such as its type, physical form, a visual element, etc. Moreover, sign-vehicles can also describe an object's pronunciation in Japanese, as is evident in kokuji of keisei, kaii-keisei, and kasha types. The kanji or its components are chosen to be the sign-vehicles due to their ability to describe an object. This ability to describe varies in terms of accuracy, ranging from a simple, seemingly arbitrary association of meanings to precise representation of a word's pronunciation.