Article
Accent Shift in Japanese and Korean
＊F
UKUI, Rei
In the historical study of Japanese and Korean accent systems, it has been assumed that a kind of accent shift took place in each language, based on regular accent correspondences among various dialects, including historical records, for each of these two languages. However, as to the direction of the accent shit, there have been opposing views. In the case of Japanese, Kindaichi (1954) and many others proposed a rightward (progressive) accent shift in a comparative study of Kyoto-type and Tokyo-type accent systems, whereas Ramsey (1979) and de Boer (2011) proposed the opposite view, a leftward shit. In the case of Korean, Ramsey (1974, 1978) proposed a letward shit but Uwano (2012) the opposite view.
he purpose of this paper is twofold: irst it will be shown that in Japanese the traditional rightward shit is preferable in explaining various peculiar accentual behaviors found in a Nairin type accent system and so it can be concluded that the rightward shit is actually occurring independently in geographically separate places. Secondly, in the case of Korean, it will be shown the letward shit is preferable in explaining accent correspondences and tones of Sino-Korean words.
he conclusion of this paper is that we have to admit opposite directions for these two languages: rightward shift for Japanese and leftward shift for Korean. Although the reason why we have opposite directions for these two languages is still unclear, the above discussion can serve as an empirical basis for considering the question of the direction of accent shit in many other languages.
Keywords:
accent shift, pitch accent, Hagiwara dialect, Middle Korean,
Sino-Korean
＊ his is a revised version of my paper presented at the symposium “Japanese and Korean accent: diachrony, reconstruction, and typology” held at ILCAA, Tokyo University of Foreign Studies on 2 July 2016. I would like to thank Ito Chiyuki for inviting me to this symposium. I am also grateful for the two anonymous reviewers for their helpful comments.
1 Introduction
2 Accent shit in Japanese 2.1 Direction of accent shit 2.2 Hida Hagiwara dialect
2.2.1 Accentual behaviors of some particles 2.2.2 Accentual patterns of infinitive
and past forms of atonic words 2.2.3 he accent of the verb ‘kakae-’ 2.2.4 Accentual features of three mora
nouns
2.2.5 Summary (Hagiwara dialect) 3 Accent shit in Korean
3.1 Basic accentual correspondences 3.2 Sino-Korean tones
3.3 When did the accent shit take place? 4 Concluding remarks
1 Introduction
This paper discusses the process of accent shift in Japanese and Korean. Both languages are known for having distinctive pitch accent or tone systems among their dialects and if we compare any two (or more) accent systems in each language, it is oten the case that they show a regular correspondence regarding the location of accent. From the viewpoint of historical linguistics, such a correspondence cannot be explained unless we posit a regular accent shit that has occurred in one or some of the dialects in question. In this paper, special attention will be paid to the direction of accent shit, since there have been opposing views regarding this issue for each language. Generally speaking, Japanese researchers tend to prefer a rightward shit but the opposite direction has been advanced by researchers outside Japan for both languages.
For Japanese, the scope of this paper is limited and I will take up only an example of the Nairin (Tokyo) type accent system in order to show that the same kind of accent shit proposed for other Nairin dialects, such as Notojima and so on, had occurred in a dialect geographically separated from the dialects discussed in previous studies. For Korean, I will discuss both modern major dialects having a distinctive pitch system and the Middle Korean (hereater abbreviated as MK).
2 Accent shit in Japanese
One of the most famous examples of accent shift in Japanese is the one found between the so-called Tokyo and Kyoto type accent systems. In many cases, accents in the Tokyo dialect appear in positions shifted one mora rightward compared to those of the corresponding Kyoto words, except for the cases beginning with a low tone in the Kyoto dialect. Therefore, it is theoretically possible to posit that either the Tokyo accents are derived from the results of a rightward accent shit, or the Kyoto accents from the results of a leftward shift. Alternatively, it is also possible to posit that both are derived from a third, common ancestor, and this is desirable for a full-scale reconstruction of the accent system of Proto-Japanese. However, as the scope of this paper is limited and for the sake of simplicity, I will only discuss the former two possibilities.
2.1 Direction of accent shit
here have been two theories concerning the direction of the accent shit in Japanese.
• Progressive shit (Rightward shit)—Kindaichi (1954), Uwano (2012) and many others • Regressive shit (Letward shit)—Ramsey (1979, 1980), de Boer (2011)
observe that the Noto dialects showed various intermediate stages that were undergoing the accent change from the Kyoto type to the Tokyo type accent system, thus supporting the rightward shit theory.
Uwano (2012: 1433) largely accepts Kindaichi’s theory of rightward shit, expressing it in the following way: “progressive shit is the most typical type of accentual change in Japanese of which we have a number of examples.”
However, it is obvious that the rightward shit alone cannot explain the fact that some words in the Tokyo type accent system are prototonic, i.e., having an accent in the first syllable that correspond to the Kyoto words beginning with a low tone. In such cases, Kindaichi (1954) proposed the “change of initial rising” (the same phenomenon is called the “word-initial pitch boost” in Uwano (2012)). he following are the typical examples of this initial rising.
(1) LL > HL LLL > HLL
LLH > HLH > HLL LLLL > HLLL
LLLH > HLLH > HLLL etc.
(Based on Uwano (2012: 1435), expression slightly changed.)
Recently, Hirako (2015) veriied that these changes are actually occurring in Noto dialects, thereby observing that the initial rising can be synchronically regarded as an overlapping of a kind of word tone {HL.}.
Contrary to these views, Ramsey (1980: 64–54) proposed the opposite direction by arguing that the Kyoto type accent system was made by shifting the position of accent in the Tokyo type accent one mora backward (letward). One of the motivations for this reversal of direction is the geographical distribution of Japanese accent systems. That is to say, the Tokyo type accent systems are distributed in geographically separated areas in the Japanese archipelago, intervened by the Kyoto type accent systems (now called the “Central” systems in accordance with the proposal made by Uwano (1987)), so that if the rightward shift advanced by Kindaichi (1954) is in fact the case, it would be difficult to explain such a geographical distribution.
In the following, I would like to show that not only the Noto dialects but also another so-called “Nairin” (inner) Tokyo type accent system that is geographically separated from the Noto areas can be said to have undergone the rightward accent shit, as proposed by Kindaichi (1954), in the relatively near past.
2.2 Hida Hagiwara dialect
Hagiwara is a small town located in the southern area of Hida where the author of this paper was born and raised1).
he historical merger pattern of the accent classes is as follows:
(2) Hida Hagiwara dialect: historical merger pattern of accent classes. Accent class
one mora noun 1 / 2 3 two mora noun 1 / 2 3 / 4 5
three mora noun 1 5b 6 7b / 2 4 / 3 5a / 7a
Because of the merger pattern for one mora noun (1/2 3), this dialect can be called a typical Nairin accent system. Phonetic pitch shapes and phonological pitch behaviors are almost the same as the Tokyo dialect.
In the following, I will present various synchronically peculiar accentual phenomena that are relevant for understanding the formation of the accent system of this dialect. he following is a summary of the topics dealt with.
• he particles ‘mo’ and ‘to’ (citation) are accented.
• The infinitive forms of unaccented verbs have a stem-final accent, and the forms with endings ‘–te’ and ‘–ta’ also have an accent and its location is determined by phonological environment.
• he irregular accent of the verb ‘kakaeru’.
• As for three mora nouns, the following observations can be made:
Class 3 nouns show regular accentual correspondence with the Heian HLL. Class 6 nouns merged with Class 1.
Class 5 and 7 nouns are split into two groups for each class (5a, 5b; 7a, 7b). Aside from the historical accent classes, there are a group of paroxytonic words
(LHL) that correspond to Kyoto and Tokyo prototonic words (HLL). Such words include relatively newer borrowings but are older than the borrowings from the early 20th century and they can serve as clues to determine when the accent change took place.
2.2.1 Accentual behaviors of some particles
In this dialect, the particles ‘mo’ (too) and ‘to’ (citation) show peculiar accentual behaviors that are rarely found in other Tokyo type accent systems. First, examples in (3) show that anything that follows the particle ‘mo’ must be lowered. Compare (3a) and (3b):
(3) Particle ‘mo’
a. Niwa ga aru. (here is a garden.)
LHH HL /μμμ μ’μ/ (μ stands for a mora, accent is represented by an apostrophe.) b. Niwa mo aru. (here is a garden, too.)
LHH LL /μμμ’ μ’μ/ ( ‘L’ is lower than the preceding low. In other words, there is a double pitch fall from mo to a and then from a to ru.)
As shown in (3b), ‘mo’ itself realizes as high but the word that follows it must be lowered even in the case of a word that begins with a high pitch. Phonologically, this particle can be represented as having an accent (/μ’/).2)
Likewise, anything that follows the particle ‘to’ (citation) must be lowered as shown in (4):
(4) Particle ‘to’ (citation) /μ’/
Iku to motta3). (I thought he would go.) LHH LLL /μμμ’ μ’ μ/
Such phenomena are not found in the Tokyo dialect and seem synchronically irregular. However, historically there can be two explanations as follows:
• Results of a rightward shit (or spreading) of the Kyoto tone of the particle with a low tone.
Niwa mo aru.
(Kyoto) HHL LH > (Hagiwara) LHH LL
/μμ_μ _μμ/ > /μμμ’ μ’ μ/ (‘_’ represents Kyoto low tone.) • A possible alternative:
In some texts of the Ruijumyōgisyō, the particle ‘mo’ is marked with the light level tone (平軽). he accent appearing on this particle might be a remnant of this.
According to the former explanation, these particles became accented because the obligatory low pitches of these particles in the Kyoto accent are transferred to the word that follows these particles as results of the rightward shit or rightward pitch spreading.
he second explanation is based on written records of some Heian period dictionaries. If we interpret the ‘light level tone’ as a falling pitch, this might be rendered into an accent in this dialect. However, this cannot explain the particle ‘to’4), and it is diicult to admit such an explanation because there are no other cases of the ‘light level tone’ realizing as an accent in this dialect.
2) When this particle follows a word ending in L, it is simply realized as L, e.g., the pitch pattern of the sentence Ame mo huru (It rains, too) realizes as HLL HL.
3) he form ‘motta’ is from ‘omotta’ (thought) with the irst syllable dropped by haplology. 4) This particle was always marked as L regardless of phonological environment (cf. Kindaichi
By the way, the existence of the phenomena described above has been known for more than 40 years. Kindaichi, in a collection of his articles on Japanese accentology, added the following supplementary note (Kindaichi (1975: 259) Note 28., translation mine):
[28] As another evidence that the second type of accent system is newer (than the irst type), we can count the fact that the class 1 disyllabic noun plus the particle ‘mo’ has an accent fall at the end of the word (○●●’) in Nara-ken Yoshino-gun Totsukawa Hiraya dialect, besides other evidence noted there. his kind of feature can also be found in Ehime-ken Ochi-gun Yoshiumi-cho Tsushima dialect, and in Gifu-ken Hagaiwara-cho
dialect according to Prof. Tsuzuku Tsuneo.（[28]ほかの乙種アクセントが新しく変化し
たことを示す証としては〔付記〕に述べた以外に，奈良県吉野郡十津川村平谷方言（中略）
は乙種方言であるが，第一類２拍名詞＋「も」の形に，カゼモ（○●●’）のように語末の
滝が現れることも数え得ると思う。このような性質は，愛媛県越智郡吉海町津島方言にも
見られ，都竹通年雄氏によると岐阜県萩原町方言にも見られるという。）
2.2.2 Accentual patterns of infinitive and past forms of atonic words
Next, we would like to consider the accentual behaviors of the infinitive5) and past forms of atonic verbs shown in (5).
(5) Ininitive and past forms of atonic verbs.6)
Predicative Ininitive Past Past followed by a noun a. kiru (wear) ki’- kita’ (wore) kita’ mo’no7) (LH LL) b. akeru (open) ake’- ake’ta (opened) ake’ta mo’no (LHL LL) c. okiru (put) oki’- o’ita (put) o’ita mo’no (HLL LL) d. yobu (call) yobi’- yonda’ (called) yonda’ mo’no (LHH LL) e. iku (go) iki’- itta’ (went) itta’ mo’no (LHH LL)
In the case of the Tokyo dialect, these forms are atonic except for some usages of ininitive forms as was demonstrated by Tsuzuku (1951). Among other so-called Tokyo type accent systems, Nairin dialects tend to show similar pitch patterns to Hagiwara dialect while most of the Chūrin dialects show pitch patterns similar to those of the Tokyo dialect.
If we compare pitch patterns of the past forms with those of the corresponding Kyoto dialect, an interesting correlation is found. First, all the Kyoto forms have an accent in the irst syllable, and unlike the Tokyo dialect, all the Hagiwara forms have an accent either on
5) ‘Ininitive’ is Samuel Martin’s term for ‘ren’yōkei (連用形)’. As can be seen in this table, all the predicates forms atonic whereas all the ininitive forms are inal accented. his can be shown by the fact that when these forms are used as the irst element of a compound verb, the pitch fall appears at the end of the irst element in this dialect, e.g., ake’hajimeru (begin to open), oki’hajimeru (begin to put) etc. However, there are exceptions: if the length of the infinitive form is one mora, this kind of pitch fall generally does not occur.
6) he data presented here are basically the same as in Tsuzuku (1951).
the irst, second or third mora as is shown in (6).
(6) Comparison of Kyoto and Hagiwara past forms. Kyoto Hagiwara a. kita (wore) μ’ μ μ μ’ σσ’ b. aketa (opened) μ’μ μ μ μ’ μ σσ’ σ c. oita (put) μ’μ μ μ’ μ μ σσ’ σ > σ’ σσ d. yonda (called) μ’μ μ μ μ μ’ σσ’
e. itta (went) μ’μ μ μ μ μ’ σσ’
In the Hagiwara dialect, the locations of accent in these forms seem irregular but it is easy to see that they are conditioned by the phonetic environment. On a moraic basis, however, they seem irregular but these can be better explained on a syllabic basis as the results of a single rightward shit, i.e., the accent of the Kyoto forms has shited one syllable rightward. he only exception to this is the case (6c) but it has to do with the fact that this dialect can have accent in a quasi diphthong with the second element /i/, such as /ai/, /oi/, /ui/ etc., but not in a syllable with a moraic nasal and obstruent. his explains the difference between (6c) on the one hand and, (6d) and (6e) on the other. Although this discussion does not logically exclude the possibility of the letward shit, the variation in the location of accent based on the phonetic environment and the existence of the word-inal accent found in (6a), (6d), and (6e), a rare phenomenon when used in forming adnominal clauses, can be more easily explained by the rightward shit hypothesis.
By the way, the forms shown in (6) are all atonic in the Tokyo dialect. It may be speculated that the Tokyo dialect preserves some characteristics of the Gairin (external) type accent system, and the accent type known as the Chūrin type, to which the Tokyo dialect is supposed to belong, is in fact an amalgam of the Nairin and Gairin types.8)
2.2.3 he accent of the verb ‘kakae-’
In most Japanese dialects, verbs and adjectives have only two patterns of accent regardless of the length of the word, so that it can be said that there are only two tones for these groups of words. However, there are some exceptions. he accent of the verb ‘kakae-’ (to hold) is an example of such exceptional cases. Kindaichi (1974: 72) called these words as the ‘kakaeru class.’
In the Hagiwara dialect, the verb ‘kakae-’ is atonic in the predicative form but the past form ‘kakaeta’ has an accent in the irst mora. his is irregular in that a normal atonic verb of the same length, ‘kasane-’ (to pile up), for example, has an accent in the third mora.
(7) Comparison of accentual patterns of the verb ‘kakae-’ in Kyoto and Hagiwara dialect. Kyoto9) Hagiwara
_kakaeru (L0) kakaeru (atonic) _kaka’eta (L2) ka’kaeta (1)
cf.
kasaneru (H0) kasaneru (atonic)
kasa’neta (H2)10) kasane’ta (usual ‘3-4’ mora class 1 accent)
This is paralleled with Kyoto ‘_kabu’toga’ vs. Hagiwara ‘ka’butoga’ (‘helmet’ plus nominative particle). According to the rightward shit theory, the pitch shape of the word /_kabu’to/ is supposed to have undergone a change like LHL > LLH > HLH > HLL in a Tokyo type accent system so that the apparently irregular pitch shape ‘ka’kaeta’ can be easily explained if we posit a four mora version of this (i.e., LHLL > LLHL > HLHL > HLLLL).
Although the above discussion does not seem directly related to the rightward shift theory, it is nevertheless important to realize that the Hagiwara dialect shows a much closer accentual correspondence to the Kyoto type accent system than the Tokyo dialect. And if we take into account the pressure of the two-tone pitch paradigm for verbs, it seems diicult to maintain such an irregular pitch shape for a long time.
2.2.4 Accentual features of three mora nouns
In this section I will discuss problems in three mora nouns for this dialect. First, the merger pattern of the historical accent classes is as follows:
(8) Hagiwara three mora nouns: 1 5b 6 7b / 2 4 / 3 5a / 7a
• Class 3: nouns belonging to this class show a regular correspondence with Myōgisyō HLL. ake’bi , cika’ra, hata’ci, saza’e, etc.
• Class 5: nouns belonging to this class are split into 5a and 5b.11) 5a: asa’hi, ito’ko, ino’ci, nami’da, suga’ta, nasu’bi, momi’ji, etc. 5b: hasira, hibasi, makura (atonic)
• Class 6: nouns belonging to this class are atonic without exception (most regular among the so-called Tokyo type dialects I am familiar with). he word-initial pitch boost did not occur for this class of nouns.
• Class 7: nouns belonging to this class are split into 7a and 7b. 7a: ka’buto, ka’iko, ku’jira, ta’yori, cu’baki, u’siro, ya’mai, etc. 7b: icigo, karasi, kusuri, tamago, tarai, etc. (atonic)
9) he Kyoto data are based on Kindaichi (1974: 72) and Nakai (2002: 213).
10) he Kyoto forms of the verb ‘kasane-’ are taken from ‘peripheral’ Kyoto accent data in Nakai (2002).
Aside from the historical classes, the following three mora nouns, all used frequently in daily life, show an interesting correspondence among the Kyoto, Tokyo and Hagiwara dialects.
(9) Selected three mora nouns showing a peculiar accentual correspondence. Kyoto Hagiwara Tokyo gloss
mi’kan mika’n mi’kan (蜜柑 mandarin orange) me’gane mega’ne me’gane (眼鏡 glasses)
ki’rei kire’i ki’rei (綺麗 pretty) kyo’nen kyone’n kyo’nen (去年 last year) sya’kusi syaku’si sya’kusi (杓子 ladle)
sya’moji syamo’ji sya’moji (しゃもじ rice scoop) zu’kin zuki’n zu’kin (頭巾 hood)
hi’noki hino’ki hi’noki (檜 cypress) bo’tan bota’n bo’tan (ボタン button) bo’tan bota’n bo’tan (牡丹 peony)
As can be seen in this table, the correspondence pattern is the following:
Kyoto—Hagiwara—Tokyo: μ’ μ μ —μ μ’ μ—μ’ μ μ
In this correspondence, it is noteworthy that the Kyoto and Tokyo dialects have the accent in the same location, and this is anomalous in itself but I will not discuss this problem in detail here. Another important thing is that all the Hagiwara forms have the accent in the second mora, suggesting that they were shited one mora rightward from the Kyoto words. In this respect, it can be said that the Hagiwara dialect shows a closer and more regular relationship with the Kyoto dialect than the Tokyo dialect, and this leads to another hypothesis that the accent change, i.e. the rightward accent shift, took place relatively recently. In order to estimate the approximate date of this change, newly created or borrowed words can be used as a way of measuring this.
First, the word ‘megane’ (glasses) is apparently newer than the words used in the historical classes, and it is diicult to imagine the people in this area used glasses several hundreds years ago. However, the 18th century poet and painter Yosa Buson (与謝蕪村, 1716–83) was said to have worn glasses, and this word was used in some of his Haiku.12) herefore, it is possible to assume that people living in this area at least knew the word no matter whether they used glasses or not.
Also, the word ‘botan (< Port. botão)’ was used in the middle of the 18th century. Ise Tadasada (伊勢貞丈, 1717 ∼ 1784）noted in his Anzai zuihicu (安斎随筆), a kind of encyclopedic work, that the word ‘botan’ is from Portuguese (vol. 31).
In Nederland, a button is called ‘konobu (< knoop)’ and in Portugal it is called ‘butan (< botão)’, which was wrongly rendered as ‘botan’ in Japan. (和蘭国にてはコノブと言
ふ、ポルトガル国にてはブタンと言ふ、それを言ひたがえて日本にてボタンと言ふなり)
If this description is correct, then the word ‘botan’ was introduced by Portuguese missionaries in the 16th century and was widely used around the 18th century.
By the way, it may be argued that, in the case of borrowings, a kind of favorite (or default) pitch patterns existed and they preferably took the pitch pattern LHL in this dialect. However, contrary to the words shown above, newer borrowings such as the following do not follow this type of correspondence, all pronounced with an HLL pitch.
(10) Borrowings newer than the words shown in (9).
ga’rasu (glass), m’isin (sewing machine), kya’becu (cabbage), ko’ruku (cork), sya’beru (shovel), da’ria (dahlia), te’suto (test), te’nisu (tennis), me’ron (melon), etc.
To sum up, it is now possible to estimate that the rightward accent shit took place in this dialect about the 18th century or later, and that corresponds to the middle or late Yedo period.
2.2.5 Summary (Hagiwara dialect)
• Various accentual phenomena found in the Hagiwara dialect show a closer relationship to an older type of the Kyoto accent system than the Tokyo dialect, and most of these phenomena can be explained if we assume the rightward accent shit.
• he accentual behaviors of the particles ‘mo’ and ‘to’ seem diicult to maintain for a long time because of their synchronically anomalous nature.
• he same can be said for accentual behaviors of ininitive and past forms of verbs and the ‘kakaeru’ class verbs because of the inluence of verbal paradigm.
• As for the question when the accent shit occurred, the three mora nouns shown in the last section give us a clue. If we follow the expression used in the title of Kindaichi (1980), it can be said that it was newer than ‘mikan’ and older than ‘misin’. hat would approximately correspond to the Middle or late Yedo period.
• What has been shown in this paper is that the Nairin type of accentual system can emerge independently in separate places. As for the other types, i.e. Chūrin and Gairin types, there still remain problems.
3 Accent shit in Korean
• Letward shit (regressive shit)—Ramsey (1974, 1978) • Rightward shit (progressive shit)—Uwano (2012)
As for the letward shit theory, Ramsey (1978: 80) stated as follows: “It can be seen from the above chart that the Kyengsang dialect has undergone the following historical rule. Kyeongsang Accent Shit (KAS) Rule. Shit the accent one syllable to the let.”
As for the rightward shit theory, Uwano (2012: 1437) states as follows: “First, the pitch fall (the lowering kernel) ‘]’ in GS shited one syllable to the right, accompanied by the rightward shit of the initial pitch rise, to yield the accent system of HG.”
3.1 Basic accentual correspondences
The following table summarizes basic accentual correspondences among Middle Korean (MK), Hamgyŏng (HG) and Kyŏngsang (KS).
(11) Basic accentual correspondences.
(H: high, L: Low, R: rising, S: any syllable, pitch of case marker, etc. is shown in parenthesis.)
No. of syllables Example (gloss) MK HG KS (north) (south) 1 nun (eye) H H H(H) H(H) mʌr (horse) L L(H) H(L) H(L) tor (stone) R H H: L or R 2 kurum (cloud) HS HL HH HH meri (head) LH LH HL HL param (wind) LL LL LH LH(L) sarʌm (person) RS HL H:H LH(H) 3 mɨcigey (rainbow) HSS HLL HHL HHL mjenɨri (daughter-in-law) LHS LHL HLL HLL minari (parsley) LLH LLH LHL LHL sonskarak (inger) LLL LLL LLH LHH samakoy (mole) RSS HLL H:HL LHH
If we compare MK and HG, the location of accent coincides except for words beginning with a rising tone, for which HG has a high pitch. he relationship between MK and KS, or HG and KS is a little complicated and can be summarized as follows:
• he initial high pitch in MK corresponds to a pitch shape like HH. in KS.
• The initial rising pitch in MK corresponds to either a high long vowel or a pitch shape like L(H). in KS.
In order to explain the last type of correspondence, it is necessary to assume that either a leftward accent shift took place in KS, as was proposed by Ramsey (1974, 1978), or a rightward shit took place in MK and HG, as was proposed by Uwano (2012).
In my opinion, however, the synchronic phonological interpretation of the MK accent system by Uwano (2012) has the following diiculties. First, words beginning with H and R in MK are treated as ‘kernelless’ (Uwano 2012: 1436). However, the most salient feature of this system is that the location of the irst H appearing in an accentual phrase is ixed and distinctive. In this sense, words beginning with an H are best regarded as prototonic. In the case of words beginning with R, the second mora of it carries the accent since R is equivalent to LH.
As for KS, it is possible to treat words having pitch patterns like HH, HHL, etc. as ‘kernelless’ or atonic, because the peculiar tonality found in these words cannot be paralleled with other accented words.
If MK words beginning with an H or R are to be treated as tonic, then the process of developing an accent from a KS atonic word is unclear according to the rightward shift theory. If, on the other hand, we take the letward shit theory, then these KS atonic words can be explained as emerging from the results of the letward shit, because if the accent at the let edge must be moved further letward, there is no place to realize an accent so that an aberrant pitch shape emerged instead.
3.2 Sino-Korean tones
It is well known that the tones of Middle Chinese are fairly regularly relected in the tones of Middle Korean Sino-Korean morphemes. According to Ito (1999), their basic correspondences can be summarized as follows:
(12) Middle Chinese and MK Sino-Korean tone correspondences. Middle Chinese Middle Korean
Level (平) L Rising (上)・Departing (去) (1) R Rising (上)・Departing (去) (2) H Entering (入) H
In MK, Sino-Korean morphemes had a fixed tone. The level tone and the entering tone regularly correspond to MK low tone and high tone respectively, and the rising and departing tones correspond to MK rising or high tone. Also, these tones are for the most part preserved in words made of Sino-Korean morphemes in modern dialects such as HG and KS. Because of this regularity, such words can be used as a criterion for determining the direction of accent shit.
in these words. In KS, however, the relationship between each Sino-Korean tone and actual pitch patterns is complicated (but it is possible to make rules to derive these pitch patterns since there are clear correspondences among MK and modern dialects, ater all).
(13) Comparison of typical Sino-Korean words.
Sino-K. MK HG KS
(North) (South) pyek (壁 wall) H H H(L) H(H) H(H) paŋ (房 room) L L L(H) H(L) H(L) pan (半 half) R R H(L) H: L, R pʌyksyeŋ (百姓 people) HS HS HL HH HH nʌyzir (来日 tomorrow) LH LH LH HL HL inkan (人間 person) LL LL LL(H) LH(L) LH(L) taycho (大棗 jujube) RS RS HL H:H LH(H)
To sum up, it can be said that Sino-Korean data go well with the letward shit theory by Ramsey (1974, 1978). That is to say, MK and HG preserve the locations of original accents whereas KS dialects shited them one syllable letward.
By the way, if we consider the process of this leftward shift a little more closely, his ‘aberrant pre-accented forms’ (Ramsey 1978: 80) in KS might be better understood by postulating an intermediate stage like the following:
(14) A revised KS accent shit. PreKS KS
*HL > *LL ( > MM) > HH *LH > HL
*LL(H) > LH(L)
*RL > H:H or LH(H)
*HLL > *LLL (> MML) > HHL *LHL > HLL
*LLH > LHL *LLL(H) > LLH(L) *RLL > H:HL or LHH
Some researchers such as Hŏ Ung (1963) transcribe modern KS forms such as HH and HHL as MM and MML, respectively. In that case, changes such as LL > (MM) > HH and LLL > (MML) > HHL can be simpliied as LL > MM and LLL > MML.
One remaining problem with this hypothesis is that the process of changes such as from PreKS *RL to modern H:H or LH(H) and from PreKS *RLL to modern H:HL or LHH.
RL > H:L > H:H RLL > H:LL > H:HL
The intermediate pitch shapes H:L and H:LL are compatible with the leftward accent shift because R is basically equivalent to LH. At the same time, the distinction between H:L, H:LL, on the one hand, and HL, HLL, on the other, is preserved because of the long vowel H:. his long vowel prevented them from merging with HL, HLL which are normal consequences of the letward shit from pitch patterns beginning with LH.
he remaining changes H:L > H:H and H:LL > H:HL are not the result of a rightward shift, but can be thought of as a kind of restructuring after the KS atonic patterns HH, HHL have been established.
As for South Kyŏngsang, the possible change process is more diicult than the above case. For now, I can only say that at some stage the original R become short and the High pitch associated with the latter half of an R was transferred to the next syllable, thereby creating a kind of tonal entity.
3.3 When did the accent shit take place?
Next, we would like to consider when the accent shit took place. If the hypothesis that the KS dialects underwent accent shit is correct as is shown in the previous section, then this change has much to do with the way pitch patterns were recorded in historical texts. If we suppose, for example, that this change took place ater the MK texts with “side dots” were recorded (i.e. after 16th century), then scholars who spoke KS dialects could have freely participated in recording MK pitches by using “side dots”. If, on the other hand, the accent change took place before the compilation of MK texts with side-dots, it would have been diicult for KS scholars to join this work.
Recently, the author of this paper reported in Fukui (2015) that most of the MK texts with ‘side dots’ were written by people from Kyŏnggido (京畿道) (including the capital Hansŏng漢城), Ch’ungch’ŏngdo (忠清道) and Chŏllado (全羅道). Very few people from Kyŏngsangdo (慶尚道) were involved in recording MK accents by using ‘side dots’. he most prominent Kyŏngsangdo scholar in the 16th century was Yi Hwang (李滉), but he did not participate in making such texts.
Based on such a distribution of MK text writers, it can be concluded that people in Kyŏnggido, Ch’ungch’ŏngdo and Chŏllado had the same or a mutually intelligible accentual system in the MK period, whereas people in Kyŏngsangdo dialects might have had a diferent accentual system at that time. his suggests that the accent shit in the KS dialects took place before the Middile Korean period.
4 Concluding remarks
accent change, such as synchronically irregular accentual behaviors and borrowings, we have opposite directions of accent shit in these two languages:
Japanese: Rightward shit. (Kindaichi (1954) and Uwano (2012)) Korean: Letward shit. (Ramsey (1974, 1978))
Inccidentally, each case is supported by examples of borrowings including Sino-Japanese (Kindaichi (1980)) and Sino-Korean. However, we are still not sure where the diference of direction came from.
References
de Boer, Elizabeth. 2011. “he split of the Tokyo type tone system into a number of subtypes, and what this can tell us about the tone system of proto-Japanese.” Japanese / Korean linguistics 20.
Fukui, Rei. 2013. Explorations in Korean historical phonology. Tokyo: Sanseidō.
Fukui, Rei. 2015. “Chūsei kankokugo no bōten o meguru ikutsuka no kihontekina kadai.” Gengo kenkyū 148: 61–80. Nihon gengo gakkai.
Hirako, Tatsuya. 2015. “Notojima shohōgen ni okeru akusento no henka.” Nihongo no kenkyū 11(1): 18–35. Nihongo gakkai.
Hŏ, Ung. 1963. Chungse kugŏ yŏngu. Seoul: Chŏngŭm-sa.
Ito, Chiyuki. 1999. “Chūki chōsengo no aklusento taikei.” Gengo kenkyū 116: 97–143. Nihon gengo gakkai. Ito, Chiyuki. 2007. Chōsen kanjion kenkyū. Tokyo: Kyūko shoin.
Kindaichi, Haruhiko. 1954. “Tōzai ryō akusento no chigai ga dekiru made.” Bungaku, 22(8): 63–84. (Reprinted in Kindaichi, 1975.)
Kindaichi, Haruhiko. 1964. Shizakōshiki no kenkyū. Tokyo: Sanseido. Kindaichi, Haruhiko. 1975. Nihon no hōgen. Tokyo: Kyōiku shuppan.
Kindaichi, Haruhiko. 1980. “Miso yori wa atarashiku cha yori wa hurui. Akusento kara mita nihonsogo to jiongo.” Gengo 9(4): 88–98. Tokyo: Taishūkan shoten.
Nakai, Yukihiko. 2002. Keihan-kei akusento jiten. Tokyo: Bensei shuppan.
Ramsey, S. Robert. 1974. “Hamkyeng kyengsang yang pangen.uy ACCENT yenkwu.” Kwuk-e hak 2: 105–132. Ramsey, S. Robert. 1978. Accent and Morphology in Korean Dialects. Seoul: Tower Press.
Ramsey, S. Robert. 1979. “he old Kyōto dialect and the historical development of Japanese accent.” Harvard Journal of Asiatic Studies 39: 157–175
Ramsey, S. Robert. 1980. “Nihongo akusento no rekishiteki henka.” Gengo 9(2): 64–76. Tokyo: Taishūkan shoten.
Sakurai, Shigeharu. 1965. “Joshi akusento no shiteki kōsatsu.” Kokugo kokubun 34(2): 36–61. Sakurai, Shigeharu. 1975. Kodai kokugo akusentoshi ronkō. Tokyo: Ōhūsha.
Tsuzuku, Tsuneo. 1951. “Dōshino ren’yōkei to akusento.” Kokugo akusento ronsō (Kishio Terakawa et al. eds.), 385–412, Tokyo: Hōsei daigaku shuppankai.
Uwano, Zendo. 1987. “Nihon hondo shohōgen akusento no keihu to bunpu (2).” Transactions of the Japan Academy 42(1): 15–70.
Uwano, Zendo. 2012. “hree types of accent kernels in Japanese.” Lingua 122: 1415–1440.
