Respectus philologicus eISSN 2335-2388
2019, vol. 36(41), p. 85–98 DOI: http://dx.doi.org/10.15388/RESPECTUS.2019.36.41.25

Emotions and Images in Li Qingzhao’s Ci Poetry: Quantitative Analysis

Hanna Dashchenko
Oles’ Honchar Dnipro National University
Gagarin ave., 72/1202, 49000 Dnipro, Ukraine
Email: annadashchenko78@gmail.com
ORCID iD: http://orcid.org/0000-0003-3432-3679
Research interests: Classical Chinese poetry, women’s poetry of the Song dynasty

Summary. This article deals with ci written by the outstanding Chinese poetess of the Song dynasty Li Qingzhao (李清照, 1084–1155?) and her poetic style Yi’an ti (易安體). The empirical material for this research is the extant 60 ci of Li Qingzhao which are presented in one of the most famous collections compiled by Xu Peijun (徐培均). Having analyzed these ci I have found several correlations concerning her poetic style. Firstly, in Li Qingzhao’s poetry ci form is positively related to ci rank: the longer ci, the higher its chance of being perceived as an “outstanding” one by commentators and poets. Secondly, the way of expressing emotions is one of the main elements that make Li Qingzhao’s long ci (zhongdiao 中調 and changdiao 長調) to be perceived as the “outstanding” ones by commentators and poets and her short ci (xiaoling 小令) as nothing very distinctive. Thirdly, negatively marked emotions are dominant in Li Qingzhao’s ci, but in most cases the poetess did not mention the reason of the sorrow of the lyrical persona or the persona this emotion was connected with. Forthly, there is a strong correlation between ci form, ci rank and types of floral images in her poetry.

Keywords: Li Qingzhao; the Song dynasty; emotions; images; quantitative analysis.

Submitted 11/05/2019 / Accepted 23/07/2019
Įteikta 2019 05 11 / Priimta 2019 07 23
Copyright © 2019 Hanna Dashchenko Published by Vilnius University Press. This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Introduction

This pioneering research is a counter-argument to a widespread skepticism about applicability of quantitative approach to medieval Chinese poetry. Such deep distrust of any quantitative line of arguments has several important implications, and one of them is reducing the modern researches on Chinese poetry to translation practices and the purely hermeneutical interpretations of already existing commentaries[1]. My research is aimed at determining the limits and opportunities of quantitative approach while studying ci written by Li Qingzhao (李清照, 1084–1155?) and her poetic style Yi’an ti (易安體). The main purpose is to find out the correlation between four variables of Yi’an style: 1. ci form distinguished by the number of characters; 2. ci rank according to “outstanding” / “mediocre” scale; 3. floral images; 4. emotions expressed in ci (sorrow, regret, sadness, etc.).

The empirical material is 60 ci of Li Qingzhao which are presented in one of the most famous collections of her works compiled by Xu Peijun (徐培均 2009). My choice of this collection is determined by two reasons. Firstly, it presents the largest number of Li Qingzhao’s ci from all modern collections (except 柯宝成 2009). It allows to get the highest level of validity and reliability while verifying research hypotheses. Secondly, commentary on Li Qingzhao’s ci is divided into four parts. We can find three of them in most collections of ci[2], but the fourth one – Attachment (附) – is unique. In this part Xu Peijun presents ci written by two Qing poets – Wang Shizhen (王士禛, 1634–1711) and Peng Songyu (彭孫遹, 1631–1700) – on Li Qingzhao’s rhymes. Thus this part of Xu Peijun’s commentary is of special interest to my research as it allows to reveal what Li Qingzhao’s ci were deserved to be imitated and, consequently, considered by other poets as the “outstanding” ones[3].

Research design

It is necessary to define the key terms of the research.

ci form or 詞體. Ci are traditionally divided into three forms: xiaoling (小令), zhongdiao (中調) and changdiao (長調), where the number of characters is the key difference between them (王兆鹏 et al. 2003: 7; 夏承焘 et al. 2001: 31–32). So, xiaoling ranges from 14 to 58 characters, zhongdiao contains from 59 to 90 characters and changdiao varies from 91 to 240 characters[4].

ci rank. I suppose that the more “outstanding” ci is or is thought to be, the more poets and commentators find it unusual and evaluate it by writing notes or comments on it. Consequently, the number of the comments on the “outstanding” ci in the poetry collection is larger. I divide all Li Qingzhao’s ci into two groups (“outstanding” and “mediocre”) depending on the length of commentaries. Inverted commas just show the conventional character of these terms because by ci rank I mean the perception of a certain ci as an “outstanding” and “mediocre” one by different commentators and poets. It therefore does not concern what Li Qingzhao’s ci really are, but what their perception by commentators, poets and scholars was and how it changed over the centuries.

The length of commentaries on 60 ci written by Li Qingzhao in Xu Peijun’s collection is 70,271 characters, i.e. the length of an average commentary on one ci is 1,171 characters. The length of 19 commentaries is much greater, so I think these ci were considered as those that deserve more attention from the commentators and poets (so-called “outstanding” ci). In general, the length of these 19 commentaries is 57.63%, i.e. the average length of a commentary on an “outstanding” ci is 2,131 characters. According to the selected criteria the other ci can be considered as those that do not attract commentators’ attention (so-called “mediocre” ci). In general, the length of these 41 commentaries is 42.37%, i.e. the average length of a commentary on a “mediocre” ci is 726 characters.

floral images are the most specific in Li Qingzhao’s poetry. In the research “images” mean “literal images” usually defined as “especially concrete and involve little or no extension of the obvious meaning of the words used to express them” (Bonn 2010: 81). I divide all floral images into four groups:

1) a nominate plant (without any details of its parts such as flowers, petals, leaves, etc.): mei (梅) in ci to the tune “浣溪沙” (晚風庭院落初);

2) the parts of a nominate plant (the parts of a certain plant such as flowers, buds, leaves, branches, etc.): flowers of mei (梅花) in ci to the tune “菩薩蠻” (梅花鬢上殘);

3) the parts of an innominate plant (the parts of the plant without stating what plant it is): branch (枝) in ci to the tune “減字木蘭花” (買得一春欲放);

4) the superordinate concepts such as花, 草 and 树, when, for instance, 花 does not mean the part of a plant but a kind of a plant as opposed to another one. E.g.: a flower (花) in ci to the tune “漁家傲” (此不與群比).

I do not include the characters formally relating to one of the abovementioned groups if they are a part of a well-established concept or idiom which are not directly connected to the plants (e.g., 芝蘭 in ci to the tune “新荷葉” is a part of a phrase 芝蘭玉樹 meaning young men’s good conduct or “喻佳子弟”).

emotions. I prefer to say about five ways of expressing emotions in Li Qingzhao’s ci, but in this research I focus only on the first three of them:

1) directly named emotions, e.g. sorrow (愁) in ci to the tune “武陵春” (載不動、 許多);

2) apophatically described emotions. I mean the situation when emotions are expressed by negation of a certain emotion, e.g. Li Qingzhao in ci to the tune “如夢令” wrote 興盡晚回舟. The specific character of this way of expressing emotions is that we can say what emotion she did not express (it is not a joy), but at the same time we cannot say exactly what the very emotion she expressed (sadness, sorrow, etc.);

3) contextually defined emotions. I mean the situation when Li Qingzhao described emotions using the characters情, 情緒, 心情, 情懷, 情意, etc., but the expressed emotions can be defined only by context. For instance, in ci to the tune “菩萨蛮” she wrote夹衫乍著心情好, thus expressing her lyrical persona’s joy of spring’s arrival though the way of expressing this emotion is neither directly named nor apophatically described.

4) emotions expressed by universal triggers, and 5) emotions expressed by cultural triggers.

Hypotheses

Finding out the correlation between four variables implies formulating the relevant hypotheses.

My first hypothesis is based on the statement that poets give preference to different poetic genres or to different forms of genre due to their natural abilities and / or being inspired by the literary traditions. Thus their poems turn out to be successful or not, “outstanding” or “mediocre”, popular or relatively unknown. Ronald Egan thinks in a similar way when he compares two outstanding Song ci poets Zhou Bangyan (周邦彥, 1056–1121) and Yan Jidao (晏幾道, 1030?–1106?) and connects their style, ci form, dominant traditions and ci rank (Egan 2006: 330):

Yan Jidao was still content to stick to the Short Song, which had always been the dominant form in literati song lyrics. The vast majority of Yan Jidao’s works (86 percent) are in this form. Zhou Bangyan concentrated instead on the Long Song. Fully 55 percent of his songs, and virtually all of his celebrated pieces, are in this form.

Consequently, my first hypothesis is ci form must be connected to ci rank, though the type of this correlation is not entirely clear. In order to verify this hypothesis I have made two tables where ci form and ci rank (as well as types of commentaries in Table 2) are related with a certain unit of measurement (number of ci or number of characters in commentaries).

The Table 1 gives the data on the distribution of Li Qingzhao’s ci depending on ci form (xiaoling, zhongdiao and changdiao) and ci rank (“outstanding” and “mediocre”).

Table 1. Li Qingzhao’s ci: form and rank

Ci form

Total number of ci

Xiaoling

Zhongdiao

Changdiao

Ci rank

“Outstanding” ci

4

7

8

19

“Mediocre” ci

32

8

1

41

Total number of ci

36

15

9

60

According to the Table 1 Li Qingzhao preferred xiaoling over the other forms: 36 ci (60%) are written in this form. Then zhongdiao – 15 ci (25%) and changdiao – 9 ci (15%) go. But ci rank has little with the frequency of using a certain ci form. In particular, only 4 xiaoling (6.67%) can be considered as “outstanding” according to the selected criteria. As to zhongdiao and changdiao the proportion of “outstanding” ci in these two forms is extremely high: 7 of 15 zhongdiao (46.67%) and 8 of 9 changdiao (88.89%). Thus, I state that the longer ci, the higher its chance of being perceived as an “outstanding” one by commentators and poets and vice versa.

This conclusion is reinforced by statistical analysis. The result for chi-square test derived from Table 1 is 22,21 and it indicates that there is linear association between two variables with a probability of 0,99. To measure the strength of a relationship and its direction Tschuprow’s T and Cramer’s V coefficient and Pearson’s contingency coefficient were used: the former is 0,61 (high) and the latter is -0,99 (very high). In other words, ci form is positively related to ci rank for Li Qingzhao’s “outstanding” poems and they are negatively related when it deals with “mediocre” ci.

Table 2. Li Qingzhao’s ci: form, rank and types of commentaries

Types of
commentaries

Ci form

Total (%)

Xiaoling

Zhongdiao

Changdiao

Ci rank

“Outstanding”
ci

Records of collation

2.07

4.44

3.52

10.03

Commentary on text

3.00

8.96

13.33

25.29

Collected
commentaries

5.92

3.15

11.89

20.96

Attachment

0.23

0.32

0.80

1.35

Total (%)

11.22

16.87

29.54

57.63

“Mediocre”
ci

Records of collation

9.10

2.16

0.17

11.43

Commentary on text

19.56

4.72

1.32

25.60

Collected
commentaries

4.44

0.25

0

4.69

Attachment

0.55

0.10

0

0.65

Total (%)

33.65

7.23

1.49

42.37

All ci (%)

44.87

24.10

31.03

100.00

The Table 2 shows the number of characters percentage depending on ci form, ci rank and four types of commentaries in Xu Peijun’s collection. The given data in whole verify the abovementioned correlations between ci form and ci rank. In particular, the other unit of measurement (the number of characters in commentaries) shows that the portion of “outstanding” ci correlates with ci form: xiaoling is 11.22%, zhongdiao is 16.87% and changdiao is 29.54%. It is remarkable that the rise in the proportion of “outstanding” changdiao to zhongdiao is virtually equal to the relevant figure in the Table 1: 1.75 and 1.90 correspondingly.

This was also confirmed by the statistics. The result for chi-square test derived from Table 2 is 39,00 indicating a linear association between two variables with a probability of 0,99. Tschuprow’s T and Cramer’s V coefficient is 0,62 (high) and Pearson’s contingency coefficient is –0,84 (very high). Thus, as in the case of Table 1, ci form is positively related to ci rank for Li Qingzhao’s “outstanding” poems and they are negatively related for “mediocre” ci.

If we look at the types of commentaries, we will also see that ci form and ci rank are negatively related to such types of commentaries as Commentary on text and Attachment: the longer ci, the higher the proportion in these commentaries on “outstanding” ci and the lower the proportion in these commentaries on “mediocre” ci. Besides, two ci forms (zhongdiao and changdiao) and ci rank are negatively related to such type of commentaries as Collected commentaries.

The greatest difference between “outstanding” and “mediocre” ci can be seen in two types of commentaries: Collected commentaries and Attachment. The length of commentary on 19 “outstanding” ci in Attachment is twice as much as the one on 41 “mediocre” ci: 1.35% and 0.65% correspondingly. If we take only zhongdiao and changdiao into consideration, this ratio is much more significant: 1.12% and 0.1% correspondingly. The longest commentary on “outstanding” ci in Attachment is in changdiao (0.80%), the shortest one is in xiaoling (0.23%). As to “mediocre” ci these figures are directly opposed: the longest commentary in Attachment is in xiaoling (0.55%) and the shortest one is in changdiao (0%).

As to Collected commentaries the figures are much more convincing. If we compare “outstanding” and “mediocre” ci in the whole, the length of such type of commentaries on “outstanding” ci is almost 21%, that is 4.5 times as much as the relevant figure concerning the “mediocre” ci (4.69%). If we look only at zhongdiao and changdiao (15 of 19 “outstanding” ci), this correlation is much more significant: 15.04% vs 0.25%.

So, I state that ci form is positively related to ci rank, and ci form is negatively related to ci rank and such types of commentaries as Commentary on text and Attachment (and to a lesser degree Collected commentaries). But it is still unclear what elements make Li Qingzhao’s zhongdiao and changdiao to be perceived as the “outstanding” ones and her xiaoling as nothing very distinctive. It is necessary to reveal the features of Li Qingzhao’s ci being dominant in changdiao, prevailing in zhongdiao and infrequent in xiaoling. As poetry deals with emotions and specific images the next two hypotheses concern the ways of expressing emotions and the peculiar features of the images.

My second hypothesis is based on Ronald Egan’s research on ci written by Zhou Bangyan. Having analyzed all his ci, the scholar comes to the conclusion that the long ci gave the poet

the space to explore the emotional and psychological complexities of the male experience of love, bringing more depth and texture to his composition about this experience than would be possible in the shorter form (Egan 2006: 330).

At the same time Ronald Egan connects “more depth and texture” to Zhou Bangyan’s composition with representing “the intimate details of romantic assignations” in ci (Egan 2006: 339), “incorporating lines from Tang and earlier shi poetry into his compositions” that made his ci difficult to understand (Egan 2006: 341) and describing nature as the complement and counterpart to the mood of lyrical persona (Egan 2006: 343). In other words, Ronald Egan sees the advantages of the long ci (changdiao) in a strictly extensive and quantitative terms, i.e. using of new stylistic devices, the appropriation of ideas, words or images from other sources, etc.

While not denying Ronald Egan’s idea, I think that it needs modifications when we talk about Li Qingzhao. First of all, it is due to her own understanding the specific features of such poetic genre, as ci. In “Essay on ci” («詞論»), first theoretical treatise on ci, Li Qingzhao was highly critical to her predecessors and contemporaries because their ci were closely connected to shi. Such close connection was unacceptable for her by many reasons. The most important of them are the lack of emotions permitted in shi and restriction to express the emotions explicitly[5]. As Li Qingzhao was focused on the problems of ci demarcation and legitimation their specific thematic focus meant, at first, a gradual shift to the very expression of different emotions and then a move from implicitly towards explicitly expressed emotions.

So, my hypothesis is the following: ci form and the way of expressing emotions are positively related: the longer ci, the more frequent Li Qingzhao used directly named emotions and the more seldom the other types of emotions, and vice versa.

At first sight this hypothesis is illogical, as we expect to find a lot of directly named emotions in short ci (xiaoling) as it is like the expectation to see more key words in a shorter message. But the connection between shi and ci can be traced even on formal level, as xiaoling with the length of 14–58 characters is almost equal to two most widespread shi forms – jueju (絕句) with 20 or 28 characters and lüshi (律詩) with 40 or 56 characters. It meant some pressure from traditional poetic canon, in particular, early ci are similar to these two shi forms not only by the number of characters, but also by implicitly expressed emotions and other traditional poetic devices[6] though they are different by theme[7]. At the same time using longer ci (zhongdiao and changdiao) might mean for Li Qingzhao having more freedom from traditional poetic canon and possibility to stress the explicitly expressed emotions.

In order to verify this hypothesis I calculate the proportion of three ways of expressing emotions depending on ci form and ci rank (see Table 3). The frequency of directly named emotions increases from xiaoling to changdiao both in “outstanding” and “mediocre” ci. It is quite predictable that the frequency of directly named emotions in changdiao is the highest one (3.25). This is also true to apophatically described and contextually defined emotions except that the frequency of contextually defined emotions in “outstanding” changdiao is lower than in “mediocre” ones: 0.63 and 1.00 correspondingly. So, I think that the abovementioned hypothesis is verified.

Table 3. Li Qingzhao’s ci: form, rank and ways of expressing emotions

Ways of expressing
emotions

Ci form

Total (ratio)

Xiaoling

Zhongdiao

Changdiao

Ci rank

“Outstanding”
ci

Directly named
emotions

1

1.43

3.25

2.11

Apophatically
described emotions

0.25

0

1.13

0.11

Contextually defined
emotions

0

0.57

0.63

0.47

Total (ratio)

1.25

2.00

4.00

2.68

“Mediocre”
ci

Directly named
emotions

1.22

1.88

3.00

1.39

Apophatically
described emotions

0.03

0.13

1.00

0.07

Contextually defined
emotions

0.31

0.63

1.00

0.39

Total (ratio)

1.56

2.63

5.00

1.85

All ci (ratio)

1.53

2.33

4.11

2.12

At the same time the correlation between ci form, ci rank and the ways of expressing emotions is not so convincing. On the one hand, the frequency of each way of expressing emotions in “outstanding” ci is higher than in “mediocre” ones. On the other hand, the differences are not as large as they could be. Firstly, this may be partly due to the lack of strict criteria to distinguish what ci can be considered the “outstanding” or “mediocre” one. Secondly, it can be the result of including doubtful ci among “mediocre” ones. If we exclude them, the total proportion of directly named emotions in “mediocre” ci is coming down from 1.39 to 1.19. Thirdly, according to the selected criteria there is no difference between “outstanding” or “mediocre” ci. At last, it can be more important what the very emotions are expressed than what the ways of their expression are.

I think that the last reason is the most relevant one. That is why I found the proportion of six directly named emotions (sorrow, loneliness[8], regret, worry, sadness and grief) accounting for 86.6% of this group of emotions[9] in Li Qingzhao’s ci (see Table 4).

The relevant data show the obvious correlation between ci form, ci rank and six directly named emotions. In particular, the specific feature of “outstanding” ci is the intensity of expressing such emotions as sorrow and grief, and, to a lesser degree, sadness, loneliness and worry. For instance, sorrow is mentioned more than twice as often in “outstanding” ci as in “mediocre” ci (0.89 and 0.41 correspondingly). Besides that expressing of these six emotions in “outstanding” and “mediocre” ci depends on ci form and the correlation between them is rather strong. The proportion of six emotions in “outstanding” ci increases smoothly from xiaoling to changdiao while in “mediocre” ci the greatest proportion is in xiaoling and the smallest proportion is in changdiao. These data support the hypothesis that there is a correlation between ci form, ci rank and six directly named emotions.

This was also confirmed by the statistical data showing moderate correlation. In particular, Tschuprow’s T and Cramer’s V coefficient between ci form and directly named emotions is 0,33 for “outstanding” ci and 0,36 for “mediocre” ci. Moderate correlations in this case can probably be explained by the fact that only one way of expressing emotions was taken into consideration.

Table 4. Li Qingzhao’s ci: form, rank and kinds of directly named emotions

Kinds of
emotions

Ci form

Total (ratio)

Xiaoling

Zhongdiao

Changdiao

Ci rank

“Outstanding” ci

Sorrow

0.11

0.32

0.47

0.89

Loneliness

0

0

0.16

0.16

Regret

0

0.05

0

0.05

Worry

0.05

0.05

0.11

0.21

Sadness

0

0.05

0.16

0.21

Grief

0

0.05

0.21

0.26

Total (ratio)

0.16

0.53

1.11

1.79

“Mediocre” ci

Sorrow

0.27

0.12

0.02

0.41

Loneliness

0.05

0.05

0.05

0.15

Regret

0.15

0.10

0

0.24

Worry

0.12

0.05

0

0.17

Sadness

0.12

0.05

0

0.17

Grief

0.07

0

0

0.07

Total (ratio)

0.78

0.37

0.07

1.22

All ci (ratio)

0.58

0.42

0.4

1.40

For more detailed interpretation of the abovementioned correlation we can have a closer look at the characters used by Li Qingzhao for describing the emotional state of her lyrical persona. In order to identify the specific features of these characters I am going to depart from the traditional practice of creating the semantic field of a certain emotion and use the classification of emotions peculiar to Chinese culture (Santangelo 2003: 206–225). It will also allow to verify to a certain degree the numerous remarks about emotional depth in Li Qingzhao’s ci (Galik 1996: 63; Hsu 1994: 73; Yao 2012: 115; Idema et al. 2004: 239; 包晓华 2011: 110) and their emotional variety “ranging from the melancholy to the erotic” (Larsen 2012: 53). In particular, according to this classification six abovementioned emotions are negatively marked (Santangelo 2003: 249). In other words, negatively marked emotions are dominant in Li Qingzhao’s ci, so her ci are associated with sorrow, loneliness, regret, worry, sadness and grief.

The next step is to examine thoroughly such emotion as sorrow, the most significant one in Li Qingzhao’s ci. In most cases for describing this emotion the poetess used a character 愁 (17.53% of all 97 directly named emotions). In her ci sorrow is described as something deep (濃愁), endless (閒愁) or fresh (新愁); it can be of great quantity (許多愁) and it can solidify (愁凝); sorrow can be as long as a thousand threads (愁千 縷) or it can linger all day long (愁永晝); sorrow makes a person suffer (愁損).

Paolo Santangelo points that 愁

includes elements from both the “unsatisfactory affects (sadness-regret-shame) complex”, the “negative projections (fear-fright-suspicion-worry) complex”, as well as the “positive expectations and interactions (love-desire-hope) complex” (Santangelo 2003: 212).

If we look at Li Qingzhao’s ci with this character, we will see a lack of the last complex. In most cases the poetess did not mention the reason of the sorrow or the person this emotion is usually connected with. It is rather unexpected because Li Qingzhao’s ci traditionally are considered as autobiographical ones (Egan 2013: 105–109; Idema 2004: 238), and the emotions expressed in them are connected with her husband Zhao Mingcheng (趙明誠, 1081–1129) (Egan 2013: 108–109; Djao 2010: 106–108; 陈祖美 2003: 50–51; 赵晓辉 2005: 38; 徐北文 2005: 56–57). The weakness of this widespread opinion can be explained by gender stereotypes according to which woman can feel sorrow only for her beloved one. Thus it could be supposed that using the character 愁 Li Qingzhao expressed her existential sorrow generated by the inability to realize her personal potential in the medieval patriarchal society (see: Dashchenko 2013).

My next step is to reveal the correlation between ci form, ci rank and types of floral images. I expect a clear distinction in frequency of using four types of floral images between two groups: “outstanding” and “mediocre” ci and xiaoling vs zhongdiao vs changdiao. The Table 5 shows the data concerning the proportion of four kinds of floral images depending on ci form and ci rank.

According to the Table 5 “outstanding” ci differ from “mediocre” ones in frequency of their using in the decreasing order a nominate plant, the parts of an innominate plant and the superordinate concepts. In “outstanding” ci a nominate plant is used more than twice as often as in “mediocre” ones (1.00 vs 0.49). Almost the same is true to the superordinate concepts (0.42 vs 0.29). But the distinguish feature of “mediocre” ci is the frequency of using the other two types of floral images: the parts of a nominate plant (0.61 vs 0.21) and the parts of an innominate plant (0.73 vs 0.53). Besides of that there are significant differences between ci form and ci rank in the frequency of using such types of floral images as a nominate plant and the superordinate concepts: it increases smoothly from xiaoling to changdiao in “outstanding”ci but in “mediocre” ci the situation is reversed. And the differences between xiaoling and changdiao are considerable. It should also be mentioned that we can see all four types of floral images in “mediocre” xiaoling, the sharp decrease of their using in zhongdiao and even more significant fall (up to complete absence) in changdiao.

Table 5. Li Qingzhao’s ci: form, rank and types of floral images

Types of floral images

Ci form

Total
(ratio)

Xiaoling

Zhongdiao

Changdiao

Ci rank

“Outstanding”
ci

Nominate plant

0.16

0.21

0.63

1.00

Parts of a nominate plant

0

0.16

0.05

0.21

Parts of an innominate
plant

0.21

0.16

0.16

0.53

Superordinate concepts

0.05

0.16

0.21

0.42

Total (ratio)

0.42

0.68

1.05

2.16

“Mediocre”
ci

Nominate plant

0.32

0.15

0.02

0.49

Parts of a nominate plant

0.46

0.15

0

0.61

Parts of an innominate
plant

0.49

0.22

0.02

0.73

Superordinate concepts

0.19

0.10

0

0.29

Total (ratio)

1.46

0.61

0.05

2.12

All ci (ratio)

1.13

0.63

0.37

2.13

In a whole, there are 128 floral images in 60 ci of Li Qingzhao but she used different types of floral images in three ci forms. So, more than a half of 128 floral images (53.13%) are in xiaoling, and in 36 xiaoling the poetess preferred to use such type of floral images as the parts of an innominate plant (35.29% of all floral images in this form). In most cases she described fallen flowers or buds (45.83%) and their fragrance (20.83%). The parts of a nominate plant make 27.94% of all floral images in xiaoling. Li Qingzhao gave priority to flowers or buds of a certain plant (78.95%) and in most cases this image deals with mei (梅). Such type as a nominate plant makes 23.53% of all floral images in this group and mei is the most frequently mentioned (37.5%).

In 15 zhongdiao we can find 38 floral images (29.69%). As in the case of xiaoling, Li Qingzhao preferred to describe the parts of an innominate plant (31.58% of all floral images in zhongdiao). But if in xiaoling she described fallen flowers or buds, in zhongdiao her preference is given to flowers and branches (41.67% and 33.33% correspondingly). It should be noted that describing a nominate plant Li Qingzhao didn’t prefer any of them as it is the case of mei in xiaoling. If we talk about such type as the parts of a nominate plant, she described different parts of a willow (branches, catkins or floss).

There are 22 floral images (17.19%) in 9 changdiao. Unlike xiaoling and zhongdiao the most frequently mentioned type of floral images in this form is a nominate plant (59.09% of all floral images in changdiao). There is no any preference in usage of certain image as it is in zhongdiao. But in this form Li Qingzhao mentioned not only real plants but also mythical ones (e.g., 階蓂or扶桑). Moreover, the diversity of mentioned kinds (species) in changdiao is much higher than in the other forms. If we do not take into consideration three kinds of a nominate plant (梅, 柳 and 梧桐), as they are used in all ci forms, we will see that Li Qingzhao mentioned four kinds (海棠, 芭蕉樹, 菊 / 黃花 and 竹) in xiaoling, three kinds (桃, 李 and 松) in zhongdiao and six kinds (黄花, 松, 酴醾, 階蓂, 扶桑 and 椿) in changdiao. In this ci form Li Qingzhao gave preference not to flowers, buds or branches, but to the fragrance of an innominate plant.

These trends with some reservations remain while dividing into “outstanding” and “mediocre” ci. The first difference is the lack obvious preferences on certain floral image in “outstanding” ci. At the same time image of mei is the most frequently mentioned one (45%) in “mediocre” ci. The second difference is that the most of floral images in “mediocre” ci belong to the parts of an innominate plant (34.09%), while in “outstanding” ci Li Qingzhao preferred to describe a nominate plant (47.5%). Thus, I state that there is a correlation between ci form, ci rank and types of floral images.

This was also confirmed by the statistics, because Pearson’s contingency coefficient between ci form and ci rank is –0,976 showing a very strong negative correlation.

Conclusion

In this article I found out several correlations concerning Li Qingzhao’s ci and her poetic style Yi’an ti. Firstly, in Li Qingzhao’s works ci form is positively related to ci rank: the longer ci, the higher its chance of being perceived as an “outstanding” one by commentators and poets. Secondly, the way of expressing emotions is one of the main elements that make Li Qingzhao’s long ci (zhongdiao and changdiao) to be perceived as the “outstanding” ones and her short ci (xiaoling) as nothing very distinctive. I think that using longer ci (zhongdiao and changdiao) might mean for Li Qingzhao having more freedom from traditional poetic canon and a possibility to stress the explicitly expressed emotions. Moreover, negatively marked emotions are dominant in Li Qingzhao’s ci, but in most cases the poetess did not mention the reason of the sorrow or the person this emotion was connected with. I state that using, in particular, the character 愁 Li Qingzhao expressed not her sorrow for her beloved husband but rather existential sorrow generated by the inability to realize her personal potential in the medieval patriarchal society. Finally, there is a correlation between ci form, ci rank and types of floral images. For example, “outstanding” ci differ from “mediocre” ones in frequency of their using of a nominate plant and usage not only real plants but also mythical ones.

This research shows the great possibility of quantitative approach while studying Chinese medieval poetry. At the same time, its results imply further attempts to operationalize new concepts and verify a range of new hypotheses as well as to connect quantitative analysis with qualitative one in more sophisticated ways. It will allow to understand more precisely the correlation between images (not only floral ones), various ways of expressing emotions (including emotions expressed by universal triggers and cultural triggers), the length of ci, its perceived quality and intersection between traditional poetic canon and individual talent.

Acknowledgments

This article is based on a paper presented at the Second Conference on Middle Period Chinese Humanities 14–17 September 2017 at Leiden University (the Netherlands). I would like to thank my two reviewers (Horst W. Huber and Jingyao Zhou) and discussion facilitators (Michael Fuller, Benjamin Ridgway, Jeffrey Moser and Lee Sukhee) for their valuable comments and suggestions. I am also very grateful to Chiang Ching-kuo Foundation for International Scholarly Exchange (Taiwan) and European Association for Chinese Studies for their support (Library Travel Grant, 2017). I am also thankful to Sergiy Shkrebets (Kharkiv, Ukraine) for providing the invaluable assistance as a statistical consultant (SPSS Statistics 19 was used).

Sources

Bonn, J., 2010. A Comprehensive Dictionary of Literature. Delhi: Abhishek.

Chang, K., 1980. The evolution of Chinese tz’u poetry from late T’ang to Northern Sung. Princeton: Princeton University Press. https://doi.org/10.2307/2054954.

Dashchenko 2013. – Дащенко, А., 2013. Личностное измерение стиля И-ань: ценностные установки Ли Цинчжао. Respectus philologicus, 24 (29), pp. 72–86. Available at: http://www.rephi.knf.vu.lt/images/24_29/2_2%20Respectus%202013%2024(29)%20%20Online%20Issn%20Dashchenko.pdf [Accessed 21 April 2019].

Dashchenko 2016. – Дащенко, A., 2016. «Рассуждения о цы» («詞論») Ли Цинчжао. Respectus philologicus, 30 (35), pp. 35–44. Available at: http://www.rephi.knf.vu.lt/images/30_35/Dashchenko.pdf. [Accessed 21 April 2019]. https://doi.org/10.15388/respectus.2016.30.35.04.

Djao, W., 2010. A Blossom Like No Other: Li Qingzhao. Toronto: Ginger Post Inc.

Egan, R., 2006. The Problem of Beauty: Aesthetic Thought and Pursuits in Northern Song Dynasty China. Cambridge: Harvard University Press. https://doi.org/10.2307/j.ctt1tm7frm.

Egan, R., 2013. The Burden of Female Talent: the Poet Li Qingzhao and Her History in China. Cambridge: Harvard University Asia Center. https://doi.org/10.2307/j.ctt1dnn96n.

Galik, M., 1996. Melancholy in Europe and in China: Some Observation of a Student of Intercultural Process. Asian and African studies, 5 (1), pp. 50–68.

Hsu, K., 1994. The Poems of Li Qingzhao (1084–1141). In: Images of women in Chinese literature. Ed. Y. Li. Indianapolis: University of Indianapolis Press, pp. 73–95.

Hu, P., 1966. Li Ch’ing-chao. NY: Twayne publishers, Inc.

Idema, W., Grant, B., 2004. The Red Brush: Writing Women of Imperial China. Cambridge: Harvard University Asia Center. https://doi.org/10.2307/j.ctt1tg5kw2.

Larsen, J., 2012. Li Qingzhao. In: Critical Survey of Poetry Asian Poets. Ed. R. Reisman. Hackensack: Salem Press, pp. 53–60.

Lin, S., 1978. The Transformation of the Chinese Lyrical Tradition: Chiang K’uei and Southern Sung Tz’u Poetry. Princeton: Princeton University Press. https://doi.org/10.1017/s0041977x00146373.

Pannam, C., 2009. Music From A Jade Flute. The Ci Poems of Li Qingzhao. Melbourne: Hybrid Publishers.

Pian, R., 2003. Sonq Dynasty Musical Sources and Their Interpretations. Hong Kong: The Chinese University Press.

Rydholm, L., 1998. In Search of the Generic Identity of Ci Poetry. Doctoral dissertation. Stockholm: Stockholm University.

Samei, M., 2004. Gendered Persona and Poetic Voice: The Abandoned Woman in Early Chinese Song Lyrics. Lanham: Lexington Books.

Santangelo, P., 2003. Sentimental Education in Chinese History: An Interdisciplinary Textual Research on Ming and Qing Sources. Leiden: Brill.

Sun, C., 2011. The Poetics of Repetition in English and Chinese Lyric Poetry. Chicago: University of Chicago Press. https://doi.org/10.1086/671941.

Yao, D., 2012. Chinese Literature. Cambridge: Cambridge University Press.

包晓华, 2011。 蒙古族女诗人那逊兰保与李清照之比较。 大连民族学院学报, 2, 110–112, 122。

陈祖美, 1995。李清照评传 。 南京:南京大学出版社。

陈祖美, 2003。李清照词新释辑评。 北京:中国书店。

陳祖美, 2004。李清照 。臺北市:知書房出版集團。

柯宝成, 2009。李清照全集:汇评本 。武汉:崇文书局。

施议对, 1989。词与音乐关系研究。北京:中国社会科学出版社。

施議對, 2006。李清照全閱讀。香港:三聯書店有限公司。

王兆鹏, 刘尊明, 2003。宋词大辞典 。南京:凤凰出版社。

夏承焘,吴熊和, 2001。读词常识 。北京:中华书局。

徐北文, 2005。李清照全集评注。 济南:济南出版社出版。

徐培均, 2009。李清照集箋注。上海:上海古籍出版社。

杨合林, 2001。 李清照集。长沙:岳麓书社。

赵晓辉, 2005。中国古典诗词精品赏读丛书:李清照。北京:五洲传播出版社。


[1] Ronald Egan’s aversion (Egan 2013: 104) of quantitative attribution of Li Qingzhao’s doubtful ci is the best example of such position. He rejects any new way of solving this problem following some assumptions: ci is a short form using just a few dozen characters with a limited range of subjects and highly conventionalized language and feelings; that is why it generated a large number of undistinguishable stylistic imitations. But such position, in fact, turns a modern scholar just into an interpreter of the established opinions on Li Qingzhao’s ci given by the very first commentators and does not allow any criticism of these commentators if their views on ci are similar.

[2] These three parts are: 1) Records of collation (校記) is the comparison of characters and phrases used in Li Qingzhao’s ci from different editions; 2) Commentary on text (箋注) is the explanation of characters, phrases and images with reference to the other poems or ancient texts; 3) Collected commentaries (匯評) is the commentaries on ci starting from Song dynasty to the modern times.

[3] Xu Peijun separates the doubtful ci from the authentic ones and puts them into a chapter titled as “存疑辨證”. It is also allows to verify the hypotheses concerning the peculiar features of Yi’an style. The editors of some collections also put the doubtful ci into a separate chapter (徐北文 2005; 柯宝成 2009; 杨合林 2001), but the principles of this separation are unclear. As Ronald Egan aptly notes, when they distinguish between authentic and doubtful ci “the basis for that distinction remains ad hoc and impressionistic. It does not spring from an analytical consideration of publication history and the problem it entails” (Egan 2013: 105). Furthermore, ci being doubtful in one collection can be stated as the authentic ones in the other. For instance, Ke Baocheng considers ci to the tune “浣溪沙” doubtful (柯宝成 2009: 92–93) while two other editors mark it as an authentic one (徐培均2009: 70–72; 徐北文 2005: 93–95).

[4] Some scholars use the other classifications. For instance, ci are sometimes divided into only two kinds or sub-genres: 小令and 慢詞 (Hu 1966: 19). The former “contains not more than 62 characters” while the latter “ranges roughly from 70 to 240 characters” (Chang 1980: 212). Sometimes researchers mention four kinds (令, 引, 近 and 慢) according to their affinity to the four major “melody categories” (Lin 1978: 107; Rydholm 1998: 45; 夏承焘 et al. 2001: 31). There is also a classification of nine ci forms (法曲, 大曲, 慢曲, 引, 近, 序子, 三台, 纏令, 諸宮調) which is also directly connected to the kinds of ci melodies (施议对 1989: 189; Rydholm 1998: 46). The latter two classifications are irrelevant to my research as they are not based on the text of ci but on the musical component of ci. The loss of the most original melodies (Pian 2003: 7, 33; Pannam 2009: 10) does not allow to accept such classifications as the reliable ground while searching the correlations between the abovementioned variables.

[5] I examined part of them in the commentary on Li Qingzhao’s “Essay on ci” (Dashchenko 2016).

[6] It is also should be noted that early ci were written mainly by men but in female voice (Samei 2004: 1, 29; Idema et al. 2004: 220). Leaving aside the matter of male and female background in ci, I point out the hypothetic influence of gender stereotypes. In other words, in patriarchal society in medieval China men could be less likely to express emotions explicitly than women.

[7] This feature of early ci where the love theme was dominant is considered by many experts as the main difference from shi: ci is classified as “feminine” and therefore “weak” being an opposition to shi regarded as “masculine” (Rydholm 1998: 154).

[8] Though “loneliness” is not an emotion, but rather a state, I include it among other emotions, as Li Qingzhao wrote about it very often (9.28%) and in almost 90% “loneliness” is directly connected to such emotions as sorrow and sadness. In some cases researchers also put loneliness among other negative emotions (e.g. Santangelo 2003: 300).

[9] Researchers have repeatedly pointed out that Li Qingzhao focused on expressing negative, specifically, sad emotions (Yao 2012: 115; Idema et al. 2004: 239; 包晓华 2011: 110). In certain respects it corresponds to the dominant opinion in Chinese literature that “verses of sorrow are easy to perfect as poetry lends itself to expressing sadness” (Djao 2010: 36).