Politeness Pressure on Grammar: The Case of First and Second Person Pronouns and Address Terms in Korean

Grammar is constantly emergent as an aggregate whole of discourse tendencies that are present in language use between interlocutors, hence the notion ‘emergent grammar’ (Hopper 1987). These tendencies are formed by diverse discursive needs, including the need to signal politeness, which is assumed to be universal (Brown and Levinson 1987). This need is particularly important in Korean, in which politeness is highly grammaticalized, i.e., the politeness marking is not only a pragmatic but grammatical issue. The two areas where the speaker’s decision is most clearly visible are the choice of sentence-enders, modulated up to six levels, and the choice of personal reference, e.g., pronouns and address terms. This study is a diachronic investigation of the personal reference system in Korean, exploring the effect of pressure of politeness. Despite the high level of grammaticalization of politeness marking, the personal reference system is a highly unstable paradigm, i.e., it has not undergone a high level of ‘paradigmaticization’ (Lehmann 1995 [1982]). Since personal reference terms are highly variable, the speakers often avoid using them for fear of the addressee perceiving that the choice is of insufficient honorification or that the very act of using reference terms is impolite when they could be omitted. Furthermore, personal reference terms with the [+Honorific] feature constantly deteriorate through frequent use. Therefore, a look into Korean reference terms shows that [+Honorific] terms are constantly innovated to upgrade the diminishing honorification effect and the first-person reference terms are constantly innovated to strengthen the [+Humiliative] meaning.


Introduction
The system of personal pronouns in Korean has not received much attention among linguists, except for discussions on the syntactic enabling factors for reflexive pronouns. There is a body of literature addressing synchronic states of pronominal systems in the history of Korean. Notable exceptions include Lee (1978), Kim (1995Kim ( , 2001, Kim (1998), Song (2002), Heine and Song (2012), and . A diachronic investigation of the system, however, reveals interesting aspects of grammaticalization of lexical expressions into personal pronouns as well as on-going fluctuation within the paradigm.
in Korean. Most prominently, Korean sentences, properly ended, should be marked with the speech level, depending on the relative social hierarchy between the speaker and the hearer, known as the honorification system, marking the speaker's deference to the addressee (see Brown 2015). This system is complex and variable with four to seven different levels of honorification. It includes corollary rules of deference-marking, and pervasively applies to first-, second-and third-person references. Another area to which honorification applies involves the relative hierarchy of the referent as compared to the speaker, known as the subject honorification or referent honorification. This is marked with the morphological marker -시 (-si) on the verb.
In Korean politeness and honorification constitute the two major determinants of speech levels. Politeness is marked with -요 (-yo) at the ultimate position, whereas honorification has many different grammatical manifestations (see 2.2 below).

Personal Pronouns in Korean: The Status Quo
One of the notable characteristics in Korean is that the pronominal system as a grammatical paradigm is not well developed in the sense that the extent of paradigmaticization is limited and the paradigm boundary with the lexicon is not rigid. Thus, there are many members in the paradigm and the internal cohesion among the members is necessarily weak.
As noted above, the levels of honorification are fine-grained and strongly tied to the grammatical system. Honorification, for its inherent nature of other-directedness, is not applicable to 1SG, as it is an act of referring to someone with deferential attitude (cf. Nam and Ko 1993). Dishonorification, as an act of referring to someone with humiliative attitude (cf. Koo 2004), can be applied to all persons, and self-dishonorification is a strategy for politeness, as is thought to be universal in politeness research (Levinson and Brown 1987). Dishonorification of one discourse participant/referent may result in relative honorification of the other(s). Honorification in pronominal reference is grammatically encoded by choosing (i) a polite form for 1SG (Humiliative/ Humble; HUM, e.g. 저 (ce)), (ii) an honorific form for other persons (e.g. 그대 (kutay), 당신 (tangsin), 댁 (tayk), etc.), (iii) a non-honorific form for a person other than the addressee (when the addressee is a social superior to the person being referred to), known as honorification suppression (e.g. 형 (hyeng) 'older brother' instead of the HONsuffixed 형님 (hyeng-nim), when speaking to his father about him), or (iv) a combination of any or all of them.
Another issue closely related to the pronominal system involves terms of address, the latter often regarded as encompassing the former. Since the pronominal reference system is not well developed in Korean, nominal address terms are widely used in lieu of pronouns (see 3.3 below). The personal pronominal system in Modern Korean (MoK; 20 th ~ 21 st centuries; note that 21 st century Korean is also called PDK) is fluid and shows variable perceptions of individual researchers. For instance, Song (2002) uses a three-level system, consisting of High HON(orification), Middle HON, and Low HON, and a more fine-grained level distinction is found in Sohn (1999: 207), as exemplified in Table 1 (note that D stands for a demonstrative, speaker-proximal 이 (i) 'this', speaker-distal 그 (ku) 'that', and mutually-distal 저 (ce) 'that').

D-어른 (D-elun)
referent's status is higher; highly HON As noted above, the level distinction is subject to considerable variation by individuals, and the distinction in Table 1, elegant as it is, is not intuitively straightforward for many speakers. The fluidity of the system is well illustrated by the fact that Sohn (1999: 207) indicates that elusin for deferential 2SG is "rare", which might be true at the time of writing (i.e., 1990s). However, it is now one of the most frequently used deferential 2SG in PDK three decades thereafter (see below). This state of affairs of fluidity is reflected in the systems proposed by other researchers, who, instead of preset levels, use contexts, which may seem ad hoc, e.g., Nam and Ko (1993 81-83), as shown in Table 2 (in which D also stands for a demonstrative).
As noted above, the inventory of personal pronouns varies across time as well as by researcher. According to a corpus search based on a section of the 21 st Century Sejong Corpus (805,606 words, spoken, dated from 2001-2015), the frequency of each pronominal form in the descending order is 1SG 나 ( are not attested, which suggest that they are becoming, or already have become, obsolete in PDK.

A historical survey of personal pronouns
The contemporary writing system of Korean, known as Hankul (or Hangeul), was invented in 1443, the landmark year that divides Early Middle Korean (EMK; 10 th c. ~ Mid-15 th c.) and Late Middle Korean (LMK; Mid-15 th c. ~ 16 th c.). With the invention of the alphabetic script Hankul, unambiguous interpretation of the historical Korean data was made possible. Before its historic invention, Chinese characters were used for their sound value (phonogram) or semantic value (semantogram), and thus interpretation of the pre-Hankul data has been often controversial. Thus, the historical depth of the data is relatively shallow and the available data sources are not rich in styles and genres for complete analyses. However, the available data sources provide sufficient evidence to support the following analysis.
An investigation of the pronouns attested in the history of Korean reveals that there have been interesting changes, whereby the overall pronominal system has become increasingly complex through time. Individual pronouns cannot be discussed in detail (some select forms are discussed in 3.2 and 4.4), but overall representative forms attested in history can be presented as Table 3, in which the forms in bold denote newly emerged forms (excluding the variants); the underscored forms denote those with the HON feature; and the forms with an asterisk*, those with the HUM feature. D denotes a demonstrative. The rows are intended to reflect the relative degree of the HON feature, i.e., the higher the row in the cell, the greater the degree of HON 3 . Based on the development of pronouns in history we will discuss some issues from the grammaticalization perspective, such as lexical sources, shifted reference, the indistinct lexis-grammar boundary, and grammaticalization parameters. 3 In EMoK, diverse Sino-Korean 1SG terms came into the system such as 소인 (soin) 'small person', 소자 (soca) 'small son, small person', 소생 (sosayng) 'small student, small person', (all involving the Sino-Korean 소 (so) 'small'), and 불초 (pwulcho) 'one who falls short of the parents' virtues, an incompetent person', etc. (see 4.4 for more discussion). The first occurrences of 소인 (soin) are 24 tokens in two texts dating from c. 1517, i.e., LMK, in the form of syozin and 小人 in Chinese characters. But since they seem to have been interchangeable with 나 (na), it seems that soin lacked the HUM feature in LMK and acquired the HUM meaning only in EMoK (Kim 2001: 12-13). In MoK and PDK, numerous nominals, prominently kinship terms and status nouns, are used in lieu of pronouns (see 3.3 and 4.1 for more discussion).
Исследование вежливости и невежливости в глобальном контексте There are two noteworthy aspects concerning the sources in Korean. The first is that many words of the Chinese origin are recruited into the pronominal paradigm (see also 4.3). The other is that diverse kinship and status nouns are used to such an extent that the boundary between the lexis and the grammar is indistinct, i.e., such kinship and status nouns are still used as fully lexical nouns and instances of use of such forms cannot be easily distinguished between lexical usage and pronominal usage. We now turn to a discussion of this latter issue.

Indistinct Lexis-Grammar Boundary
As indicated in the preceding exposition, the pronominal system in Korean is fluid and nouns constitute the most common source category of pronoun grammaticalization, a state of affairs in line with the so-called "the noun-to-pronoun channel" (Heine and Song 2010: 122). Numerous referring terms with diverse levels of honorification are used in MoK and PDK, most of which are full-fledged nouns, and most of those nouns are kinship terms and status nouns. Only a small number of defective nouns are used, e.g., 님 (nim) 'honorable person', 녁 (nyek) 'side', 곳 (kos) 'place', 것 (kes) 'thing', and 쪽 (ccok) 'side'. The following is the list of such nominal forms used for personal reference in MoK and PDK, some of which have fallen into disuse 7 : (3) a. 2SG (general): 선생님 (sensayngnim) 'teacher' (for any adult, mostly for male), 사장님 (sacangnim) 'company president' (for adult male), 사모님 (samonim) 'teacher's wife' (for adult woman), 학생 (haksayng) 'student' (for youth), 총각 (chongkak) 'bachelor' (for young male), 처녀 (chenye) 'maiden, virgin' (for young woman), 언니 (enni) 'older sister' (for young female, often in service encounters), 이모 (imo) 'maternal aunt' (for adult woman, often in service encounter), 어머니/어머님 (emeni(m)) 'mother' (for aged woman), 아버님 (apenim) 'father' (for aged man), 아줌마 (acwumma) 'aunt' (for a married woman, often avoidable), 할머니 (halmeni) 'grandmother' (for an old woman, often avoidable), 할아버지 (halapeci) 'grandfather' (for an old man, often avoidable), 고객님 (kokayknim) 'honorable patron' (for a client), N-님 (N-nim) 'honorable [name]' (for anyone, mostly in service encounter), 그쪽 (kuccok) 'that side' (for anyone), 그편 (kuphyen) 'that part' (for anyone), 그곳 (kukos) 'that place' (for anyone), 거기 (keki) 'that place' (for anyone) .. 'offspring' (for male, Pejorative); fossilized D-nouns of the 그 (ku) 'that' origin, e.g., 그쪽 (kuccok) 'that side/direction' (for anyone), 그편 (kuphyen) 'that part/side' (for anyone), 그곳 (kukos) 'that place' (for anyone), 거기 (keki) 'that place' (for anyone) ... (3), many regular nouns are used for person reference and many of these terms form complex nominals in combination with demonstratives for 3SG reference. Such complex forms exhibit differential degrees of internal fusion, i.e., a process of morpho-syntactic bonding between the words in the phrase, in which some phonological reduction may also occur. Korean orthographic rules require interlexemic spacing, according to which a phrase consisting of a demonstrative modifier and a head noun must include a space between the two words. However, the spacing is variably regulated in orthography and variably written by individuals, even though certain forms invariably occur without a space, which indicates that they have undergone grammaticalization to a great extent. For instance, 3SG 그 사람 (ku salam) 'the person' should be written with an interlexical space according to the orthographic rules, as shown here, but many Koreans write the phrase without a space, i.e. 그사람 (kusalam), thinking that it is no longer a phrase denoting 'the person' but a monolexemic pronoun denoting 's/he'. Most phrasal nominals used for person reference are placed somewhere on the continuum between the two polar extremes, which suggests that the levels of grammaticalization of these nominals are variable. This state of affairs is schematically presented as (4) with some representative forms:
The parameter of extension, i.e., using a form in a different context, is usually actualized through 'context-induced reinterpretation' (Heine et al. 1991). For instance, an addressee interprets RFL ce 'self' as referring to the speaker himself/herself 'I', based on the contextual and situational context, leading to the functional extension of ce RFL to 저 (ce) RFL and 1SG. The extension pattern is evident in Figure 1, a modified version of Table 3 with arrows added to indicate the referential extensions. As shown in Figure 1, the pronouns for 1SG and 2SG extended from RFL, e.g., 저 (ce) to 1SG, 당신 (tangsin) to 2SG and 3SG, 자네 (caney) to 2SG, and 자기 (caki) to 2SG, retain their RFL usage, thus these are the instances of true extension rather than shift or switch-reference. Furthermore, there are instances of extension observed with nominal-based pronouns. For instance, 소인 (soin) 'small person' to 1SG, 그대 (kutay) 'that place' to 2SG, 임자 (imca) 'owner' to 2SG, etc. are all instances of extension since the form with the original meaning are still in use (thus 'divergence ' Hopper 1991). This extension is observed in most, though not all, nominals listed in (3)   The parameter of desemanticization (also called 'attrition', Lehmann 1995Lehmann [1982; 'semantic bleaching ', Sweetser 1988;'generalization', Bybee et al. 1994) is also attested in Korean. For instance, expressions making use of 'house', 'body', 'place', etc. have lost such lexical meanings and now they simply denote the persons metonymically associated with the place, who are significant in the discourse situation.
Decategorialization refers to the loss in morphosyntactic properties characteristic of lexical forms. This parameter is also observed by many forms of personal reference grammaticalizing in Korean. For instance, some of the forms involving a demonstrative, e.g., 그대 (kutay) 2SG (< 그 (ku) 'that' 대 (tay) 'place'), have developed such strong internal cohesion that they cannot be interrupted by a modifier, thus *그 가까운 대 (ku kakkawun tay) [that near place] (only interpretable non-pronominally as 'the place nearby', in which case tay also changes to tey) but 가까운 그대 (kakkawun kutay) 'intimate you'. However, since many nominal-based pronouns are still in the periphrastic stage, the loss of the primary category properties is often limited. For instance, some can take modifiers, some can function as an autonomous independent form, and many can take the PL suffix -tul to be pluralized. All these are indicative of the fact that many Korean nominal-based pronouns still carry lexical properties (see also 3.3 above for variable degrees of fusion).
Erosion refers to a loss of phonetic substance. The oldest pronouns in Korean, e.g., 1SG 나 (na), 2SG 너 (ne), 3SG 이 (i), 그 (ku), and 뎌 (tye), and RFL 저 (ce), are simple in form (e.g., monosyllabic and without a coda), and their longer counterparts have not yet been identified. Therefore, there is no way of confirming if erosion occurred to these old grams. As for the newer pronouns, phonetic erosion is either absent or minimal (note that nouns used as a personal pronoun typically lose phonetic salience and thus are rarely stressed). Again this is an indication of the low level grammaticalization of pronouns as a paradigm. Incidentally, this situation is not uncommon for a large number of pronouns across languages (Heine and Song, 2010: 120).

Discussion
In the preceding section we have seen how the paradigm of pronominal reference emerged in the history of Korean (3.1 and 3.4). We also noted that the paradigm is nearly open-ended by virtue of using diverse nominals such as kinship and status words (3.2 and 3.3). We now turn to the discussion of our major focus, addressing such questions as (i) why the pronominal system in Korean has not been well developed into a paradigm with high paradigmaticity; (ii) why 2SG is often not explicitly expressed; (iii) why the pronominal system is fluctuating in references and honorification levels; and (iv) why the pronominal system becomes increasingly complex with new forms. From these inquiries we will see how the pressure of politeness constantly pushes the language users to innovate the personal reference terms.

An "Unsystematic System"
As noted in the exposition above, the Korean pronominal system is in flux. The consequence is that numerous forms proliferate in the paradigm, which in turn makes its inventory controversial. Even though there is a limitation in the available data sources in that those texts do not use 1SG and 2SG for stylistic reasons, e.g., written narratives, religious commentaries and poems, it is clear that Korean in OK and EMK had a simple, crude pronominal system, one for each person, notably without HON forms, and a 3-way 3SG distinction by distance, i.e., with the speaker-proximal i 'this', speaker-distal ku 'that', and mutually distal ce 'that'.
One of the reasons for the absence of a strong, well-established pronominal system is the Korean discourse-pragmatic idiosyncrasy that in discourse nouns (and names) are simply repeated, without having to be replaced with pronominal forms. The following is a good example, taken from the 1447 data, one of the earliest extant Hankul documents Исследование вежливости и невежливости в глобальном контексте (the long excerpt is given without a morpheme-by-morpheme interlinear gloss, and the repeated names are given in bold) 8 : (5) seyconi sangtwusaney kasya yongkwa kwisinkwa wihAya selpephAtesita. yongkwa kwi wihAya selpephAsyami pwuthyeys nahi syelhuntwulhilesini mokwang yesuschas hAy ulyuila. pwuthyey moklyenitAlye nilAsyatAy ney kapilakwukey kaa apanimskuywa acAmanimskuywa... 'Buddha went to Mt. Sangdu and preached for the dragon and the ghosts. Preaching for the dragon and the ghosts was when Buddha's age was thirty-two, which was the In example (5), the terms for Buddha, i.e., seycon and pwuthey, are simply repeated without pronominal substitution. Seycon literally meaning 'the venerable of the world' is synonymous with Buddha, originated from the Sanskrit word, meaning 'the awakened one' (cf. Phyocwun Kwuke Taysacen, web-searchable ed. accessed August 15, 2019). This kind of practice is not specific to this text, but is a general writing style in LMK. In other words, the pronoun replacement for co-referential nouns was not a discourse convention in LMK. As a matter of fact, it is not a robust discourse-syntactic rule even in PDK.
Another reason for the absence of a well-established pronominal system is that vocatives are frequently used, and then the use of pronouns can be often dispensed with, as shown in (6) As shown in (6), when a vocative is used, the sentential subject may not occur since it is contextually recoverable. This omissibility of arguments has been a syntactic idio-syncrasy in Korean. In PDK, the most common vocatives in interaction as a replacement strategy is the kinship terms, e.g. acwumma 'aunt', acessi 'uncle', halmeni 'grandmother', halapeci 'grandfather', enni 'elder sister', oppa 'elder brother', imo 'maternal aunt'. It is to be noted that these are all upward kinship terms. Downward kinship terms are not employed for this purpose. Even though they are upward kinship terms their use is not restricted to addressing a social superior, i.e., it can be used toward a social inferior. For instance, an older client may address a younger attendant at a restaurant as enni 'elder sister'. Such kinship term usage is exemplified in the following constructed, but commonly used examples: During the MoK a large number of status nouns and titles entered the system for personal reference. Constant emergence of such nouns is an ongoing process in PDK. In PDK this is arguably the most common strategy for personal reference, in the sense that whenever a person is known to have a title the most convenient way is to address him or her with the title. This is well illustrated in the following constructed examples: emma-ka ha-ycw-ulkey mom-NOM do-BEN-PROM 'I'll do it for you. (Lit. Mom will do it for you.)' As shown in (8), the status nouns are used instead of the pronouns, and further, (8b) and (8c) involve empathy or identity shift. For this reason, emma 'mother' in (8b) is the child's mother, the addressee, whereas, emma 'mother' in (8c) is the first person, the speaker 9 . Empathy is common in speech for children but it is unidirectional, i.e., downward empathy (the mother using the child's term to the child) is acceptable whereas upward empathy (the child using her mother's term to her mother) is not 10 .

Avoidance of 2nd Person
Another important factor in the (non-)development of a well-established pronominal system in Korean is the weight of sociocultural norms with respect to positioning. Positioning in discourse, according to Davies and Harré (1990: 48), is "the discursive process whereby selves are located in conversations as observably and subjectively coherent participants in jointly produced storylines" (see also Locher 2008Locher , 2018, for discussion of the close relationship between identity construction and politeness concerns). In the context of pronominal development, the concept of positioning as proposed in Song and Heine (2016: 4) refers to "defining the social role relation between speaker and hearer". Positioning is highly complex in the Korean society along such variables as age, occupation, year of college matriculation, year of initiation/affiliation with an organization, gender, etc., many of which are not immediately identifiable. Pinpointing (or identifying) the referent explicitly is burdensome as it may violate decorum. Thus, the need for an avoidance strategy (Heine and Song 2010) is strongly felt. This situation may well have led to, or contributed to, the underdevelopment of the pronominal paradigm.
Furthermore, identifying the referent explicitly is often unnecessary because of the honorification of the verbal predicate (for sentential subjects) or sentential ending (for addressees), as shown in (9) o-si-myen malha-y come-HON-COND say-END 'If (he) comes, tell me.' ('He' being honorified, the addressee not politely treated) As shown in (9), the presence and absence of the HON verbal suffix often unambiguously identifies the subject because in a given context who would and would not be honorified is often straightforwardly clear. This system of honorification would have contributed to the non-use of explicit sentential arguments and consequently that of pronouns as well. 9 An anonymous reviewer comments that (8c) is an instance of baby-talk rather than that of empathy. But this type of identity shift is not restricted to baby-talk but is widespread. For instance, it is customary for a grandfather to ask his aged daughter about his wife by referring to her as emma 'mother' rather than nay anay 'my wife', etc. The downward empathy is a norm.
10 For similar states of affairs in Japanese, see Whitman (1999: 366) and Suzuki (1976). cf. A woman to a crying child, in Japanese, Atasi doo si-ta no? [I (girl) how do-PERF Q] 'What's wrong (little girl)?' in which atasi 'I' refers to the girl, the addressee, not the woman, the speaker.

Intra-systemic Fluctuation
Another important aspect of the Korean pronominal system is that there has been intra-systemic fluctuation (see Figure 1 in 3.4). For instance, RFL forms are often recruited for 2SG. According to Heine and Song (2010) recruiting RFL for 2SG is a crosslinguistically common strategy. They note that "[t]he speaker avoids addressing the hearer directly and portrays him as being more significant or central vis-à-vis alternative referents" (Heine and Song 2010: 128). Similarly, Shibatani (1985: 837, as cited in Heine andSong 2010: 129) notes that the connection between RFL and honorifics lies in agent defocusing.
The reference shift (as part of reference extension, see 3.4 above) is attested in multiple instances. The [저 (ce) RFL > 1SG.HUM] change, for instance, is a discoursepragmatic strategy of presenting self as a third party. In this case referential identity is established through the addressee's inference, and the distance thus created helps the form acquire the HUM meaning. Similarly, RFL tangsin developed into 2SG and 3SG. The noun 당신 (tangsin) is a Sino-Korean word denoting 'the body concerned, the applicable body'. As noted in 3.2 above, a number of Sino-Korean words are involved in the development of pronouns in Korean, and such words of Chinese origin belonged to a high register in historical times since Chinese was the script of the literati (see also 4.4 below). The RFL caki that developed into 2SG is also a Sino-Korean word. The form in the 2SG reference is for endearment address, and the development is also an instance of adopting indirectness. Indirectness between spouses may seem strange but in traditional Korean the honorific and polite forms, not intimate forms, were used between spouses, a tradition still maintained in the old generation. This indirectness between spouses is also evident in such address forms as 여보 (yepo) 'Look here!' > 'Honey, Darling' and CN-아빠/엄마 (CN-appa/emma) '[child.name]-dad/mom' > 'Honey, Darling'. Another RFL 자네 (caney) (< 'people like you' or 'of you') to 2SG.HON is also an instance of a RFL form developing into 2SG, a crosslinguistically common pattern.
Nothing displays the intra-systemic fluctuation more clearly than the pronoun 당신 (tangsin), which started out its life as a RFL, developing into 2SG and 3SG. The series of changes of the form across the pronominal references as attested in the history can be diagrammatically presented in Figure 2. Figure 2 shows that 당신 (tangsin) began its life as a RFL.HON in LMK, and this HON feature is retained in the RFL function throughout history. The feature was also inherited when it extended its function to the 3SG reference. However, after it extended its function to 2SG reference in EMoK, the HON feature became bleached in MoK in certain usages. In PDK the 2SG reference function is variegated. The feature remains only in poetic and liturgical usage. It is used with the overtone of affection between spouses, but it is also used in completely neutral contexts, i.e., 'audience-blind' contexts Rhee 2013a, Rhee andKoo 2017), such as textbooks and newspaper articles. However, the self-same form can be provocative when used between people who are not spouses 11 . For this reason, its acceptability is highly sensitive to the context in which it is used. The referential versatility is largely due to its source semantics 'the body concerned, the applicable body', because the interpretation of 'concerned' or 'applicable' varies by context. Исследование вежливости и невежливости в глобальном контексте

Emergence of Honorific and Humiliative Pronouns
As was shown in Table 3 in 3.1, the earliest pronominal system in the OK and EMK period was a simple one without any form marked with the HON feature. Then the RFL.HON forms cAkya and 당신 (tangsin) appeared in LMK. At this time 2SG.HON 그듸 (kutuy) also appeared. In the EMoK period a number of HUM forms in 1SG came into existence and so did the HON forms in 2SG. In MoK, a number of new HUM forms entered into 1SG and so did the HON forms into 2SG and 3SG. The proliferation of HON forms in 2SG and 3SG is quite noticeable in PDK.
Innovation of HON forms involves certain strategies. The most prominent one is to use the Sino-Korean form, since Sino-Korean terms in general are more highly valued than their native-Korean counterparts, e.g., the synonyms for 'an old person' 늙은이 (nulkuni) (native Korean) and 노인 (noin) (Sino-Korean) and for 'a woman' 계집 (kyeycip) (native Korean) and 여자 (yeca) /여성 (yeseng) (Sino-Korean), make a sharp contrast, with the former derogatory and the latter neutral or deferential (see Rhee 2011, Narrog et al. 2018, for more discussion). Such usage of terms of foreign origin is the characteristic of learned style used by the noblemen, and thus following the same pattern in the pronominal system is not surprising. The following is the list of some of such Sino-Korean forms that either were used in history or are being used in PDK for pronominal reference: Another strategy is to be indirect either by using metonymic references or by distancing the referent by using a form that has been in use for different references, notably RFL forms (see Table 3 for instances of RFL forms developing into 2SG and 3SG). Using the metonymic connection is the more common strategy, whereby metonyms such as 'house ', 'place', 'body', etc. are  Still another strategy is to use the Sino-Korean prefix 귀-(kwi-) that denotes 'precious', 'valuable', etc., as shown in part in (13)  Still another strategy is shown in the development of 임자 (imca) 'owner' > 2SG.HON. This is a strategy of deriving honor from ownership but since the addressee does not have the ownership, the speaker is attributing feigned ownership to the addressee. Even though Korean has only one such instance, other languages seem to have a similar pattern, e.g., Indonesian 2SG tuan from Arabic 'master', Japanese 2SG kimi from OJ 'emperor' 'lord' (Shibatani 1990: 371-372, as cited in Heine andSong 2012), English sir from the ME title of honor of a knight, a baronet, or a priest (Oxford English Dictionary, 2018 online ed.), among others 14 .
On the other hand, the simple and crude pronominal system of OK and EMK periods became a more elaborate one with the addition of HUM forms. For instance, the 1SG.HUM form ce appears in EMoK, an instance of functional extension from RFL. The speaker referring to himself or herself with RFL involves self-dishonorification and self-derogation, in the sense that the speaker is not asserting the self's presence explicitly.
Since politeness is a universal requirement in interaction (Brown and Levinson 1987), speakers often use self-derogating lexical expressions. Such self-derogating terms were prolific in the history of Korean, even though their use has declined in PDK. Some of such forms are as shown in (14)  Two interesting patterns become apparent from this state of affairs. First, the early data shows a simple and crude pronominal system, without honorification marking, but diverse forms were innovated through the passage of time. Thus, diachronically a 2SG honorifiable person was referred to as in (15). Note that OK ne was for general reference regardless of HON: Secondly, the earliest system did not have a HUM reference pronoun for 1SG, but HUM forms were innovated through the passage of time. Thus, a polite speaker making humiliative reference to the self toward an HON addressee used the pronouns as in (16) In other words, continuous innovation for novel forms occurred for 2SG.HON and 3SG.HON and 1SG.HUM. The downgrading of the HON feature is noticeable particularly with 2SG. As the strength of the HON feature of an existing HON form becomes weakened, a new form with a stronger HON feature is innovated, and when the HON feature of this form becomes weakened a new form is innovated, thus continuing the innovation cycle. On the other hand, weakening the HUM feature with 1SG leads to the innovation of new forms with stronger HUM value. The cyclicity is also observed here.
The consequence of the change in the pronominal system is obvious, as shown in Table 4. The number of rows tends to increase (except for 1SG in PDK, see below for discussion); novel forms are added at the lowest row of 1SG and at the highest row of 2SG (the novel form cells are shaded for visual conspicuity). Furthermore, existing forms tend to climb up the row in 1SG, by being pushed up by the novel forms added at the bottom, whereas in 2SG, existing forms tend to go down the row, by being pushed down by the novel forms added at the top. In this context, we can discuss the possible motivation behind this series of changes. Two factors are immediately identifiable, the routinization effect and the sociocultural effect.
Routinization has numerous effects in human life. It is well known that novelty wears out as familiarity increases. Linguistic forms tend to lose semantic content through frequent use, and forms that are intended to carry emotive forces tend to lose them through repetition (see Heine et al. 1991, Heine and Stolz 2008, Heine 2009 for discussion of role of creativity and expressivity in language change). Hopper and Traugott (2003:122) note that intensifiers are particularly susceptible to frequent renewal because the force of intensification tends to be rapidly lost. We can relate this general tendency to the frequent renewal of the HON and HUM pronominal forms. In other words, the frequent change of HON forms is due to the devaluation of HON as a result of frequent use, and the frequent change of HUM forms is due to the bleaching of HUM through repeated use. In this process of showing humiliative or honorific attitude toward the addressee and encoding such attitudes linguistically, speakers tend to 'overdo' it, since doing so is safer. This linguistically sumptuous behavior has often led to grammatical change. For instance, Koo and Rhee (2013b) show how this tendency of being 'over-polite' caused the emergence of polite imperatives from promissives. Similarly, Haiman (1994) shows repetition and ritualization are important mechanisms of grammaticalization. For this reason, linguistic forms are constantly being negotiated in discourse and thus grammar is constantly fluctuating and 'emergent' (Hopper 1987).
Исследование вежливости и невежливости в глобальном контексте The sociocultural effect is a more global factor. The social stratification of the Korean society was not rigid until Confucianism became the national religion during the Joseon Dynasty (14 th -20 th centuries). Confucianism stressed a social order based on hierarchy, and meticulous attention even to seemingly trivial matters was the preoccupation of the people. The lack of sophisticated HON and HUM pronouns until the 15 th century (i.e., OK and EMK) is not coincidental; people had been less hierarchyminded under the strong influence of egalitarian Buddhism. With Confucianism taking root in Korean society, the ruling-class intellectuals studied Confucian scriptures, written in Chinese, as their daily task from dawn to dusk. Adopting words of Chinese origin in the pronominal system mainly to encode HON is also not coincidental; literati attributed high values to terms of Chinese origin. Koo (2004: 118) shows that Koreans' special attention to hierarchy may have contributed to the grammatical and lexical coding of honorification and dishonorification. Similarly, Rhee and Koo (2014) show how the sociocultural environment influenced the prolific use of causatives and passives and how their strategic use led to their functional extension to stance-marking. Koo and Rhee (2016: 318) also show that 'honorification obsession' has influenced Korean grammar and lexis on a large scale. It can be said in general that as society became increasingly complex, so did the pronominal system in order to suit the complex social stratification.
One aspect that merits a mention in this context is the relative simplification of 1SG in PDK, only with the plain 나 (na) and the HON 저 (ce). As shown in Table 4, the numerous HUM forms of the MoK period such as 소인 (soin), 시생 (sisayng), 쇤네 (soynney), 소생 (sosayng), etc. all fell into disuse in PDK. This again seems to be the result of sociocultural influences. The long-held noble vs. common distinction in social stratification was legally abolished in 1894 as part of modernization efforts, i.e., around the beginning of MoK. People seem to have felt that the use of excessively HUM forms, translatable as 'small person', was no longer necessary. It is not at all clear, however, why a democratic and egalitarian philosophy, i.e., modernization departing from rigid Confucianism, affected the HUM forms but not HON forms. It is possible that politeness simply concerns the addressee more strongly than the speaker's self. One relevant factor in support of this hypothesis is that the notion of 'intersubjectification' (Traugott 1982, Traugott andDasher 2002) typically concerns the addressee rather than the speaker.

Summary and Conclusion
The Korean pronominal system is not well developed due to social factors and idiosyncratic traits of language use, i.e. extensive use of regular nouns and titles and pro-drop. Historically, Korean lacked HON and HUM pronouns in OK and EMK. In LMK new 2SG.HON and RFL.HONs came into existence and at a later time there occurred instances of switch references, as part of referential extension, whereby 1SG.HUM was innovated. The paradigm of pronouns has become more fine-grained with new innovations.
It is argued that as society became increasingly complex, HON pronouns and referential expressions were actively innovated throughout history by recruiting certain lexical and grammatical items such as regular nouns and demonstratives. Historical states of affairs reveal the cognitive strategies in discursive reference management as evidenced by the fact that most instances of innovation involved the upward modification for the second person pronouns and downward modification for the first person pronouns in terms of the honorification hierarchy.
This research focused on the politeness pressure on the development of 1SG and 2SG pronominal forms and the effect of sociocultural change on the system. Some recent analyses investigated the change in the address terms, e.g.  for address terms between spouses in the past 70 years, in the context of rapid sociocultural change in the post-colonial times. For better understanding of the Korean pronominal system, we need this line of microscopic research on various relation types and social variables on the one hand and more macroscopic, comprehensive studies on the pronominal system as a whole on the other.