The Translation of Classical Chinese Poetry

Ray Brownrigg


Classical Chinese poetry and, in particular, Chinese regulated poetry - also known as ‘recent style poetry’ - from the Tang dynasty offers the greatest challenge to a translator.  Not only must one deal with all the subtleties of historical and cultural context and literary allusion but there is also the challenge of the rendering of the strictly regulated form of the poetry into another language.  While most translators forgo this ultimate challenge, concentrating more on an accurate translation of the intended meaning, there is a great potential for reward from bringing the visual and aural aspects of this poetry to the Western reader.  This essay attempts to show that such greater rewards are achievable, and that the inevitable trade-off is worth it.

1. Introduction

Classical Chinese poetry is generally considered to have flourished most prominently during the Tang dynasty (618-907AD).  The particular genre of classical Chinese poetry known as regulated or ‘recent style’ poetry came to prominence during this dynasty.  This genre, along with its immediate predecessor, known as ancient or ‘old style’ poetry, influenced all Chinese poetry until the modern era, that is, for more than one thousand years.  The history of Chinese poetry is well documented (Liu & Lo, Mair 1994, Owen 1996, Minford & Lau).  The history of the translation of Classical Chinese poetry is well documented by Eliot Weinberger (1987).

The translation of poetry in general is fraught with difficulties.  In addition to the usual difficulties of translation, there are the added difficulties of trying to preserve the various poetic qualities of the original.   The famous late-19th century Chinese translator Yan Fu specified the three translation criteria of  Xin, Da and Ya (信达雅), literally translated as fidelity, comprehensibility (or fluency or expressiveness) and elegance (or quality) (Yan, Zhong).  Here fidelity refers to the original author’s meaning, comprehensibility, fluency or expressiveness refers to the target language and elegance or quality refers to the overall product of the translation.  These particular ideals have been discussed at length in the intervening century, and many other ‘theories of translation’ have been expounded.  The existence of various schools of thought on the theory and practice of translation is manifest in the frequent existence of a wide range of translations available for any particular item of literature[1].  One point that must be emphasised is that the various goals can to some extent be in conflict with one another, and so there are trade-offs to occur when performing a translation.  The resolution of these trade-offs must take into account the motives of the translator, as well as such things as the target audience.

In the context of translation of Chinese regulated poetry there is a fourth conflicting criterion which is not completely contained within Yan Fu’s three.  This is the preservation of the regulated physical form of the original.  The physical form of the original poem has no bearing on the preservation of the poet’s meaning, on the fluency of the translation, or on the elegance of the resulting poem.  If anything, the adherence to regulations in the original poem is a measure of the language skills of the original poet.  The reproduction of this skill in a translation is of course not a reflection of the original poet, but of the translator.  Nevertheless, there is no reason not to aim to reproduce this skill in a translation, thereby revealing the full glory of Tang poetry.

The reader, or more accurately the receiver, of translated poetry may be motivated by one or more of a wide range of reasons for doing so.  The two most disparate of these are reading for enjoyment and reading for academic analysis.  For the purposes of this essay, the emphasis is on reading, or listening, for enjoyment, for which purpose the technical quality of a translation is not considered to be as important as the perceived enjoyment of the receiver[2].  As a corollary, this author believes that the primary impact of a poem must occur on the first reading (or hearing, if the poem is being read aloud by somebody else).  Thus for the purposes of this essay, the ‘accessibility’ of a poem to a wide audience is more important than technical quality.  To this end, rhyme and metre play a large part.  That this should be the case is of course a personal preference.

Now considering in particular the preservation of the form of regulated verse, there are two aspects which may work in the translator’s favour.  Firstly, English is a very rich language; there often being many words which have the same or a similar meaning.  This should make it easier for the writer to conform to constraints of rhyme and metre.  Secondly, poetry that conforms to constraints will almost certainly contain some contrivances, either in the choice of word or the choice of message, and thus the need for accurate translation is reduced. Finally, considering the translation of poetry in general, poetry is already an interpretation.  It is the interpretation of the poet’s thoughts and feelings in the poet’s language of choice.  Reinterpreting accurately those thoughts and feelings in another language, while challenging, should not be impossible, as long as it is possible to determine just what the poet’s thoughts and feelings were.  This again serves to reduce the need for concentrating on a strict accuracy of translation, since the same thoughts and feelings can be expressed in different ways using different actual words.

Good translated poetry is good poetry first and a good translation second, but it is still necessary to have both a good translator and a good poet to produce a translated poem.  Nevertheless, the pre-existence of earlier attempts at translation can provide any translator with the groundwork for a new translation.

Often regarded as the best ever classical Chinese poet, or at least the best proponent of regulated verse, is Du Fu (712-770) (Owen 1981 p.183, Hung p.1).  For this reason, there are frequently many translations of Du Fu’s poems.  Thus it seems appropriate to choose Du Fu as the poet of study for this analysis.

The next section will provide some historical background to the flourishing of classical Chinese poetry and in particular the regulated verse of Du Fu during the Tang dynasty.  Following this a selection of Du Fu's regulated verse poems will be analysed in terms of the 'regulations' to which they were expected to conform.  Then the process of discovering multiple translations is described, along with some analysis of the different translations in terms of the conflicting aspects of accuracy, faithfulness, quality and form.  Finally some new translations are attempted with a view to retaining the regulated characteristics of the originals.

2. Background

2.1 Poetry

Poetry can mean different things to different people, but the English word has a Greek root meaning to create (“Poetry”).  In this context a reasonable definition of what is meant by poetry is “art created from language”.  This definition makes no assumptions about the particular language being employed, about the way in which the art is recorded nor, more importantly, about the way the ‘artistic experience’ is transmitted to the ‘receiver’.

2.2 Tang Dynasty Poetry

The development of Tang dynasty poetry had it roots in what is now known as gushi (古詩), which began to be developed in the second century AD (Frankel 1976 p.213).  From then until the 5th century gushi was the predominant form of poetry, characterized only by fixed-length lines and a fixed rhyme scheme.  The first signs of the development of the prosody rules which define regulated verse were evident in the latter half of the fifth century with the issuing by Shen Yue in 488 of his manifesto on tonal prosody (Mair & Mei 1991 p.378) (Chang “What is Jintishi?”).  By the middle of the sixth century various developmental schools of poets had worked through the unresolved issues and refined the rules (Mair & Mei 1991 p.396) to what became jintishi (近體詩), the name being used to distinguish it from gushi.  The latter was not replaced, but remained a valid poetic form even during the height of the popularity of jintishi which occurred during the Tang dynasty.

The two basic forms of jintishi are lüshi (律詩) or "regulated verse", and jueju (絕句) or "truncated verse" which are distinguished primarily by line length (eight and four lines respectively), and the jueju’s somewhat relaxed constraints on parallelism within its couplets.  These two forms are each then subdivided into their respective five-character and seven-character forms, resulting in wulü (五律), qilü (七律),  wujue (五絕), and  qijue (七絕).

2.3 Some General Comments on Translation

There are many journal papers, even complete journals, and monographs dedicated to the theory of translation.  However, the theory of translation is not the primary theme of this essay.  Here it is hoped to present an insight into a very narrow field of translation; the translation of poetry, and not just poetry but Tang dynasty poetry, and not just Tang dynasty poetry, but Tang dynasty regulated poetry written by Du Fu.  Nevertheless some general comments are still appropriate to the more specific task at hand.  Bente Elsworth writes “A translation is never a transparent pane of glass allowing a perfect view of the original.  It is always another text, a different work.” (p.3). While this may be true of translation in general, it is particularly true of literary translation.  A corollary of this is that “This [translation] process never ends because there can be no definitive translation of any one literary text.” (Bortolotti-van Loon p.105)

2.4 Translation of Poetry

It is not only a ‘school of thought’ as mentioned in the introduction that determines the balance of emphasis on conflicting ideals of accuracy, fidelity and quality.  The motives of the translator can significantly alter this balance.  The context of a translation includes aspects not only of the original text and author, but also those of the translator and the intended audience.  For example Tony Barnstone reports “Sometimes I've deviated slightly from a literal translation in order to get an effect that I believe is truer to the poet's vision. There are no fast rules …” (p. 72).  Richard Jackson goes even further, quoting Stephen Mitchell “the well known translator of Rilke” (without reference): “with great poetry, the freest translation is sometimes the most faithful.”

2.5 Translation of Poetic Form

There are definitely conflicting schools of thought on the issue of translating poetic form, or poetics in general – including rhyme, rhythm, parallelism and structure (such as fixed line length).  A good discussion of the dichotomy between form and spirit is provided by Sin-wai Chan, who nevertheless concludes “It is imperative to realize that as far as poetry translation is concerned, form cannot be reproduced.” (p.109)  Stephen Owen states “There is also no way to echo the forms of Chinese poetry and still produce translations that are accurate and readable.” (1996 p.xliv), and much earlier, William Hung had stated “I have found it difficult to cast my translation in English meter and rhyme. […] I try, therefore to convey only Tu Fu's thought and spirit, and cease to worry over form.” (p.13)  Eliot Weinberger goes further to state “Chinese prosody is largely concerned with the number of characters per line and the arrangement of tones – both of which are untranslatable.  But translators …often may be seen attempting to nurture Chinese rhyme patterns in the hostile environment of a Western language.” (1987 p.5)

However there is no lack of contrary views.  C. John Holcombe states “Free verse is hopelessly inappropriate to the regulated, highly compressed and structured nature of classical Chinese poetry.” (“Pros and Cons of the Draft”), and Arthur Cooper had earlier commented that “much of our own 'free verse’ of today seems to me closer to the Chinese fu than to either shih or tz’u.  This I think an inherent fault in attempts to translate Chinese metric verse into ' free verse' or prose.” (p.61).  Tony Barnstone, referring to the emphasis on accurate translation at the expense of rhyme and meter - a poem’s “right to sing” - comments “Too often translators have given Chinese poets the resolution powers of an electron microscope, but have cut off their ears.” (p. 75).

2.6 Translation of Du Fu

The talent of Du Fu is unquestioned; he is frequently compared to Homer or Shakespeare, yet his poetry is as yet not fully available to non-speakers of Chinese.  That he was not particularly popular in his own time, and thus much of his work has not survived to this day, is even more distressing.  Arthur Waley, in the preface to his 1946 publication Chinese Poems, states “I have indeed made many attempts to translate Li Po, Tu Fu and Su Shih; but the results have not satisfied me” (p.6).  If somebody with the experience and talent of Waley has found himself in this situation, what hopes have most of us?

Du Fu is known to have produced over one thousand[3] poems which have survived but the largest anthologies in English[4] that seem to be available are those of Florence Ayscough with around 400 translations (in two volumes) and William Hung with 374.  This represents by far a minority selection of what is available.  Other relatively prolific translators of Du Fu into English include Burton Watson, Rewi Alley and A. R. Davis, each with close to 200 translations.  

3. The Poems Chosen

The particular selection of poems for this project was chosen during the process of accumulating translations.  Essentially the poems ‘selected themselves’ by being poems for which a good selection of different translations was easy to discover.  The actual process is detailed further in Section 4.  Table 1 below lists the 11 poems chosen, with their titles in Chinese, pinyin and typical English translation.  This is followed by an analysis of these poems in terms of the regulations of jintishi.



Pinyin title

Typical English title


wàng yuè

Gazing at Taishan


yuè yè

Moonlit Night


chūn wàng

Spring Outlook


yuè yè yì shě dì

Thinking of My Brothers on a Moonlit Night


tiān mò huái lǐ bái

Thinking of Li Bai at the End of the Sky


ké zhì

A Guest Arrives


chūn yè xǐ yǔ

Welcome Rain on a Spring Night


wén guān jūn shōu hé nán hé běi

News that the Imperial Army has Recaptured North and South of the River


lǚ yè shū huái

Night Thoughts of a Traveller


dēng gāo

Climbing High


jiāng nán féng lǐ guī nián

Meeting Li Guinian South of the River

Table 1.  List of poems chosen for analysis

3.1 Analysis of Original Poems

The rules governing ‘Recent Style Poetry’ involve the three aspects of tone, rhyme and parallelism.  In general there seems to be reasonable agreement over what the rhyme and parallelism rules were, but there seems to be less general agreement amongst researchers on the rules governing tones or, more generally, prosody.

There are available several different descriptions of the prosody rules governing Regulated Poetry.  These mostly differ in minor ways from each other, but nevertheless this can affect the degree to which individual poems can be considered to be conforming to the rules.  The descriptions, summarised here in publication order, will be referred to by the authors’ names.  In order to highlight the differences between the published interpretations, the common aspects will be presented first and then the different interpretations will be presented as refinements of these basic rules.

3.1.1 Basic Rules of Recent Style Prosody

1)      All ‘regulated’ poems (lüshi or jueju) consist of four or eight lines of either five or seven characters.  All lines occur as couplets.

2)      The same rhyme is used throughout, occurring at the end of even-numbered lines and optionally at the end of the first line also.

3)      There is a caesura between the second and third characters of each line and an additional caesura between the fourth and fifth characters of each line in the seven-character form.

4)      When there are eight lines (lüshi), there is parallelism between the third and fourth lines and the fifth and sixth lines, i.e. the second and third couplets each exhibit parallelism.

5)      All characters are divided into two groups of tones, known as level and deflected tones.  These are defined in terms of the Classical Chinese tones, with Level being the original level tone, and Deflected being the original rising, falling and entering tones. 

6)      Each line follows one of only four different tone patterns.  These can be defined in terms of the seven-character line, with the five-character line omitting the first two characters (and the first caesura).  The four patterns are as follows, where / denotes a minor caesura, // denotes a major caesura, L denotes a level tone, D denotes a deflected tone and a subscript R denotes rhyme[5].

a)      L L / D D // D L LR

b)      D D / L L // D D LR

c)      D D / L L // L D D

d)      L L / D D // L L D

7)      Each set of four lines follows one of four different sets of tone patterns.  These are defined in terms of the line patterns above, namely abca, badb, cadb or dbca.  In lüshi (eight lines), the patterns are abcadbca, badbcadb, cadbcadb and dbcadbca.

3.1.2 Variations Published

The variations that can be identified from the publications of earlier researchers are almost exclusively related to the ‘liberty’ allowed in choosing the tones in some positions within lines.  The earliest found reference to the prosody of Chinese Regulated Poetry is from John Fryer in 1901 (p.xcii), who does not specify the tone rules, but as an example of “regular poetry” gives the pattern cadb above, but with the fifth and seventh tones swapped.  In association with rule 2) above, which he does state, this pattern implies the use of rhyme on a deflected tone.  While currently not considered to be common, this is not at variance with some of the much more recent descriptions such as Downer & Graham and Mair & Mei described below.

James J.Y. Liu (Liu, 1962 p.26) relaxes Rule 6 (”some liberty is allowed”) for the first, third fifth characters of seven-character lines (and by extension the first and third characters of the five-character form).  If we signify an optional tone with a dash “-” then according to Liu the four basic jintishi forms are:

-  L / -  D // -  L LR   -  D / -  L // -  D LR   -  D / -  L // -  D D    -  L / -  D // -  L LR         -  L / -  D // -  L D    -  D / -  L // -  D LR  -  D / -  L // -  D D    -  L / -  D // -  L LR

-  D / -  L // -  D LR    -  L / -  D // -  L LR   -  L / -  D // -  L D    -  D / -  L // -  D LR               -  D / -  L // -  D D    -  L / -  D // -  L LR   -  L / -  D // -  L D    -  D / -  L // -  D LR

-  D / -  L // -  D D    -  L / -  D // -  L LR   -  L / -  D // -  L D    -  D / -  L // -  D LR                -  D / -  L // -  D D    -  L / -  D // -  L LR   -  L / -  D // -  L D    -  D / -  L // -  D LR

-  L / -  D // -  L D      -  D / -  L // -  D LR   -  D / -  L // -  D D    -  L / -  D // -  L LR                 -  L / -  D // -  L D      -  D / -  L // - D LR    -  D / -  L // -  D D    -  L / -  D // -  L LR

G. B. Downer and A. C. Graham (p.147) also relax Rule 6 for the first and third characters of each seven-character line, but do not extend this to the fifth character (thus only the first character of five-character lines is affected).  Further, they relax the requirement that the rhyme be a level tone, though stating “The tone of the rhyme, [is] normally but not necessarily level.”  If the rhyme is a deflected tone, then this is achieved by swapping the tones of the fifth and seventh characters (third and fifth characters in the five-character form).  Thus as well as the four basic forms listed below, there are another four forms where the fifth and seventh tones are swapped.

-  L / -  D // D L LR  -  D / -  L // D D LR  -  D / -  L // L D D  -  L / -  D // D L LR         -  L / -  D // L L D    -  D / -  L // D D LR  -  D / -  L // L D D  -  L / -  D // D L LR

-  D / -  L // D D LR  -  L / -  D // D L LR  -  L / -  D // L L D  -  D / -  L // D D LR         -  D / -  L // L D D    -  L / -  D // D L LR  -  L / -  D // L L D  -  D / -  L // D D LR

-  D / -  L // L D D    -  L / -  D // D L LR   -  L / -  D // L L D   -  D / -  L // D D LR        -  D / -  L // L D D    -  L / -  D // D L LR   -  L / -  D // L L D   -  D / -  L // D D LR

-  L / -  D // L L D    -  D / -  L // D D LR   -  D / -  L // L D D   -  L / -  D // D L LR        -  L / -  D // L L D    -  D / -  L // D D LR   -  D / -  L // L D D   -  L / -  D // D L LR

Hans H. Frankel (pp213-215) uses the same basic structure as Downer and Graham above.

Wai-lim Yip (pp226-230) also relaxes Rule 6 for the first and third characters of the seven-character line, but not for every line.  In particular the a) lines in Rule 6) above do not have any flexibility in the tone of the third character.  Thus according to Yip the four basic forms are:

-  L / -  D // D L LR  -  D / L L // D D LR  -  D / -  L // L D D   -  L / -  D // D L LR        -  L / -  D // L L D    -  D / L L // D D LR  -  D / -  L // L D D   -  L / -  D // D L LR

-  D / L L // D D LR  -  L / -  D // D L LR  -  L / -  D // L L D   -  D / L L // D D LR        -  D / -  L // L D D    -  L / -  D // D L LR  -  L / -  D // L L D   -  D / L L // D D LR

-  D / -  L // L D D    -  L / -  D // D L LR  -  L / -  D // L L D   -  D / L L // D D LR        -  D / -  L // L D D    -  L / -  D // D L LR  -  L / -  D // L L D   -  D / L L // D D LR

-  L / -  D // L L D    -  D / L L // D D LR  -  D / -  L // L D D   -  L / -  D // D L LR        -  L / -  D // L L D    -  D / L L // D D LR  -  D / -  L // L D D   -  L / -  D // D L LR

Victor H. Mair and Tsu-lin Mei (1991 p.408) relax the tonal constraints of the first, third and fifth characters of the seven-character form[6] in the same way that Liu does but they impose a stricter requirement on the non-rhyming characters at the line ends, requiring these to “show maximum tonal differentiation among themselves”.  This phrasing is important because it is also stated (p.407) that the rhyme words may be either level or deflected (with the non-rhymed line endings belonging to the other tone category).  Thus if the rhyme characters are (presumably all the same) one of the deflected tones, the non-rhyming characters must all be the same level tone.  This would not violate the rule because there is no tonal differentiation possible with the level tone.  Further, Mair and Mei do not mention rhyme on the first line, but this is relatively easy to accommodate into their analysis consistently with the other analyses presented here.  Thus according to Mair and Mei the four basic forms are:

-  L / -  D // -  L LR  -  D / -  L // -  D LR  -  D / -  L // -  D D    -  L / -  D // -  L LR             -  L / -  D // -  L D1  -  D / -  L // -  D LR  -  D / -  L // -  D D2   -  L / -  D // -  L LR

-  D / -  L // -  D LR  -  L / -  D // -  L LR   -  L / -  D // -  L D    -  D / -  L // -  D LR               -  D / -  L // -  D D1  -  L / -  D // -  L LR  -  L / -  D // -  L D2   -  D / -  L // -  D LR

-  D / -  L // -  D D1  -  L / -  D // -  L LR  -  L / -  D // -  L D2   -  D / -  L // -  D LR                -  D / -  L // -  D Da  -  L / -  D // -  L LR  -  L / -  D // -  L Db   -  D / -  L // -  D LR

-  L / -  D // -  L D1  -  D / -  L // -  D LR  -  D / -  L // -  D D2   -  L / -  D // -  L LR                 -  L / -  D // -  L Da  -  D / -  L // - D LR   -  D / -  L // -  D Db   -  L / -  D // -  L LR

 The further four forms incorporating non-level rhyme are assumed[7] to be as follows, where D1 represents the any one of the deflected tones, but remaining the same throughout the poem.

-  L / -  D // -  L D1R -  D / -  L // -  D D1R -  D / -  L // -  D L  -  L / -  D // -  L D1R            -  L / -  D // -  L L      - D / -  L // -  D D1R -  D / -  L // -  D L  -  L / -  D // -  L D1R

-  D / -  L // -  D D1R -  L / -  D // -  L D1R  -  L / -  D // -  L L -  D / -  L // -  D D1R               -  D / -  L // -  D L     -  L / -  D // -  L D1R -  L / -  D // -  L L  -  D / -  L // -  D D1R

-  D / -  L // -  D L     -  L / -  D // -  L D1R -  L / -  D // -  L L  -  D / -  L // -  D D1R                -  D / -  L // -  D L     -  L / -  D // -  L D1R -  L / -  D // -  L L  -  D / -  L // -  D D1R

-  L / -  D // -  L L     -  D / -  L // -  D D1R -  D / -  L // -  D L  -  L / -  D // -  L D1R                 -  L / -  D // -  L L     -  D / -  L // - D D1R  -  D / -  L // -  D L  -  L / -  D // -  L D1R

Edward C. Chang relaxes Rule 6 for the first character, most third characters and some fifth characters of the seven-character form.  This produces something similar to Yip above where the a) lines of Rule 6 have no flexibility in the third character, but with some extra flexibility in the fifth character of the b) lines.  Thus according to Chang the four basic forms are:

-  L / -  D // D L LR  -  D / L L // -  D LR  -  D / -  L // L D D   -  L / -  D // D L LR        -  L / -  D // L L D    -  D / L L // -  D LR  -  D / -  L // L D D   -  L / -  D // D L LR

-  D / L L // -  D LR  -  L / -  D // D L LR   -  L / -  D // L L D   -  D / L L // -  D LR        -  D / -  L // L D D    -  L / -  D // D L LR  -  L / -  D // L L D   -  D / L L // -  D LR

-  D / -  L // L D D    -  L / -  D // D L LR  -  L / -  D // L L D   -  D / L L // -  D LR        -  D / -  L // L D D    -  L / -  D // D L LR  -  L / -  D // L L D   -  D / L L // -  D LR

-  L / -  D // L L D    -  D / L L // -  D LR  -  D / -  L // L D D   -  L / -  D // D L LR        -  L / -  D // L L D    -  D / L L // -  D LR  -  D / -  L // L D D   -  L / -  D // D L LR

David Hawkes also analyses a selection of Du Fu poems in terms of prosody, but just presents the results, without detailing the actual rules used as a basis for the analysis.  Nevertheless, those results proved to be very useful for the following analysis, since they indicated quite clearly that current pronunciation is not sufficient for such analysis.  The first[8] of our selected poems is specified by Hawkes as being gushi rather than jintishi.  This is reinforced by the significant lack of conformance to the rules regarding tones, while still preserving the structure and rhyme exhibited by gushi.  For this reason, the first poem is omitted from the analysis following.

3.1.3 Analysis Results

Initial analysis of the chosen poems was carried out using current pronunciation as an indicator of tones, the current first tone being considered equivalent to the earlier level tone, and the current second through fourth tones being considered as the earlier deflected tones.  This proved somewhat fruitless, with large numbers of apparent tone violations.  Next a similar analysis was carried out using the current first and second tones as an indicator of the earlier level tone as suggested by Edward C. Chang.  While this showed an improvement in the conformance of the poems to the rules, there were still too many apparent violations to consider the method appropriate.  In particular, the two poems described by Hawkes as “formally perfect”, namely春望 (chūn wàng) and月夜憶舍弟 (yuè yè yì shě dì), still exhibited two and three tone ‘violations’ respectively[9].  Further, the current pronunciations did not seem to provide the necessary matching rhymes for those two “formally perfect” poems.  It then became clear that the only way properly to check the tonal and rhyming compliance of the poems was to find some description of the pronunciation of the time.  A brief foray into searching for “Tang dynasty pronunciation” resulted in the discovery of the Unihan database ( which then led to Hugh M. Stimson’s monograph, which provides “Middle Chinese” pronunciation for the characters used in a large number of Tang poems. 

Using the Stimson pronunciations, the formal perfection of the two poems cited as such by Hawkes is confirmed.  Further, it seems that several others of the self-selected set are also completely compliant in terms of both tones and rhyme.  Table 2 lists the numbers of tonal non-compliances for the 10 chosen jintishi poems, according to the various rule variations set out above[10].


Title                         \  Rulelist






yuè yè






chūn wàng






yuè yè yì shě dì






tiān mò huái lǐ bái






ké zhì






chūn yè xǐ yǔ






wén guān jūn shōu hé nán hé běi






lǚ yè shū huái






dēng gāo






jiāng nán féng lǐ guī nián






Table 2.  Tone violations for selected poems.

Thus six of the nine[11] lüshi poems show perfect tonal conformance according to all the different tone rules defined above.   As far as rhyme is concerned, all the poems demonstrate perfect or near-perfect rhyme.  To illustrate what is meant by “near-perfect” rhyme, the third poem in the table above uses the rhyme words hæng, shiɛng, miæng, shræng and biæng in lines 1, 2, 4, 6 and 8 respectively.  Now while these use two different phonemes for the finals of the characters, Stimson (p. ix) states that the ɛng group of finals “could be collapsed” into the equivalent æng finals, indicating that the distinction in sound was a historical one by the time of the Tang dynasty.

The remaining aspect of regulation that needs to be checked is the parallelism of the two middle pairs of lines.  In this regard, just two of the nine lüshi do not readily pass this test.  The first of these two is 月夜 (yuè yè), which seems not to exhibit parallelism in the second couplet.  The couplet, with its literal translation, is


distant pity/love          little                 son      daughter


not       understand      remember       Chang an

where little and remember do not seem to conform in grammatical function.  However one of the features of Classical Chinese is that in general any word can acquire any grammatical function if appropriate.  Following this path, there are two possibilities to remedy the lack of conformance.  Firstly, (xiǎo) could be considered to be a verb, in which case the meaning might be belittle or reduce, neither of which seem to be appropriate.  The second is to consider () as an adjective, with the meaning memorable which is possible, but apparently not previously considered to be a likely meaning.

 The other poem not conforming to the parallelism requirement is 月夜憶舍弟 (yuè yè yì shě dì), for which the parallelism of the third couplet is questionable.  Here we have


have    brothers          all        divide  scatter


no        family              ask      die       live

where “all” and “ask” do not seem to conform, and it does not seem very easy to consider (jie, all) as a verb, or to consider (wen, ask) as a pronoun in this context.

To summarise the analysis of the eleven chosen poems, one poem turns out to be gushi, and 10 are jintishi, of which one is a jueju, and of the remaining nine, five are fully conforming lüshi.

4. Existing Translations

One of the goals of this project was to find as many different translations as possible for a selection of Du Fu poems.  It was for this reason that the poems were ‘self-selected’, as mentioned above.  The initial candidates were those poems published in Du Fu Shi Xuan (Feng).  Given this list of 167 poems, an initial scan of readily available English language anthologies was performed in an attempt to find translations.  Two primary sources were considered to be readily available.  These were the VUW library and the internet.  The VUW library electronic catalogue was very readily able to provide lists of monographs containing appropriate keywords such as “Chinese poetry” or “Tu Fu”.  This then led to the discovery of particular call numbers which were potentially fruitful, providing large numbers of relevant volumes in a single physical location.  The internet was also accessed primarily through a search process, namely the Google search web site.  Various search terms such as “Tu Fu” or “Du Fu” in combination with “poetry” and/or “translation” provided a good initial set of anthology web sites which were used to build up the numbers of translations.

During this preliminary process, no attempt was made to ensure that the translations were unique.  Further, there was a potential for some translations to be overlooked because of the way the title had been translated.  Because this author was, at this very preliminary stage, somewhat unfamiliar with any of the poems, it was only the obviously matching titles that were recognised as matches.  Apart from the similarity of title, the one feature that was mostly used to confirm the potential of a match was the number of lines in that translation, since most of the translations did match the number of lines in the original poem.   There were, however, some false positives from similarity of titles, mainly due to the existence of multiple Du Fu poems on similar subjects, such as the moon, night-time or rain.

This procedure resulted in an initial list of 16 poems which were found to have been commonly translated.   More accurately, this was a list of poems which occurred most frequently in the anthologies available, without taking uniqueness into account. This list was then at one stage extended by another 11 poems which did not appear in Du Fu Shi Xuan but were nevertheless found to be reasonably commonly translated.  Eventually though the list was pared down to a final 11 of which only one was not found in Du Fu Shi Xuan.

The next stage was to start collecting the translations together in electronic form so that the unique ones could be identified, and so that the complete list could be electronically published – one of the goals of this project.  Interestingly, this task led to a burgeoning of the numbers of translations found.  In an effort to save transcription from a printed document the procedure used was to take the first phrase or line of a translation in a monograph, and enter that as an entire phrase into the Google search engine.  In most cases, a match was found, and almost invariably when a match was found, the whole translation was available in electronic form.  Of course the chances of finding the complete translation were augmented by the fact that there were no long poems in the final eleven poems chosen.  The surprising result of this process is that frequently the match occurred as part of an on-line anthology, which provided even more translations of the chosen poems.  Further, in several cases, large selections of other translations of the same poem were discovered, usually by Chinese translators.  This explains why some of the chosen poems have more than twice the average number of translations found.

The process of searching on an opening line was not restricted just to those translations which were not already available in electronic form.  Searching for a poem which was already available on-line would lead to other web sites which provided that same translation.  Frequently some of those sites provided otherwise undiscovered translations.  These were not only alternate translations of the particular poem specified for the search, but also included further translations of other poems on the list.

One of the disadvantages of using the internet, of which every researcher should be aware, is that the quality of the information available is unknown – and certainly variable.  For this reason one must be careful to verify independently any information available from the internet, or alternatively to state explicitly that the information has not been verified.  During the process of accumulating translations as described above, a variety of different sources of error were noticed.  These include, but are not restricted to the following:

  • Poems attributed to the wrong author
  • Translations attributed to the wrong translator
  • Misspellings of poem titles
  • Missing words or lines from translations
  • Incorrectly specified pinyin
  • Incorrectly specified Chinese characters for the original poem

Some of these errors found may genuinely fall into the category of ‘typographical errors’, but the ease with which such errors can be – and indeed have been – replicated by others means that extra care should be taken whenever publishing electronically. 

As already mentioned, many of the web sites discovered were sites providing anthologies of Chinese poetry at various levels of categorisation – Chinese poetry, classical Chinese poetry, Tang poetry, or Du Fu poetry.  However there was a wide variety of types of web site encountered, such as electronic journal web sites (e.g., electronically published book sites (e.g. Burton Watson’s The Selected Poems of Du Fu at, on-line academic learning sites (e.g., on-line general literature web sites (e.g., book review sites (e.g.,  on-line poetry sites (e.g., professional web sites (e.g. Fu Sample.html), personal research web sites (e.g., web blog (personal ‘diary’) sites (e.g. and cultural interest sites (e.g. and

Table 3 below presents the final list of poems chosen and enumerates the form of each poem (see Section 2.2) and the numbers of translations, including literal translations, found for each one.




# translations

Gazing at Taishan



Moonlit Night  



Spring Outlook



Thinking of My Brothers on a Moonlit Night



Thinking of Li Bai at the End of the Sky



A Guest Arrives



Welcome Rain on a Spring Night



News that the Imperial Army has Recaptured North and South of the River



Night Thoughts of a Traveller   



Climbing High



Meeting Li Guinian South of the River



Table 3. Number of translations for each of the poems chosen

4.1. Analysis of Existing Translations

The premise of this essay is that the preservation of form during the translation of Chinese regulated poetry is a rare occurrence which need not be so rare.  In order to determine just how rare this occurrence is, the discovered translations need to be analysed in some objective way to provide a measure of the conformance to the various aspects of poetic form.  At the same time though there must be a similar measure of how ‘approachable’ the translations are in terms of other, perhaps conflicting, measures. Fortunately, since conformance of Chinese poetry to the regulations of jintishi is easily measured, it is also reasonably easy to measure translations of jintishi against similar regulations.  Of course it is not possible to use all the prosodic regulations when checking the English translations, since English does not have the concept of tones[12] and the English word is far different from a Chinese character.  Nevertheless there are some of the qualities of Chinese regulated poetry which can be exhibited by an English translation.  In particular, the concepts of fixed line-length, parallelism and rhyme are all able to be carried over into English.  Further the caesura and couplet structures of jintishi are also available to a translation.

Considering all these possibilities, the translations need to be measured against the following criteria of poetic form, which correspond to the first four basic rules of jintishi prosody.

1)      Each poem consists of four or eight lines of equal length.  All lines occur as couplets.

2)      The same rhyme is used throughout, occurring at the end of even-numbered lines and optionally at the end of the first line also.

3)      There is a caesura dividing each line approximately in two and an additional caesura in the longer form which divides the first part in two.

4)      When there are eight lines (lüshi), there is parallelism between the third and fourth lines and the fifth and sixth lines, i.e. the second and third couplets each exhibit parallelism.

The remaining rules are all related to the very strict order of tones, which are not appropriate to the English language.

Some of the above four rules consist of multiple elements, each of which may be determined independently.  Taking this into account, each poem was scored out of a maximum of seven points, being:

1)      matching the number of lines of the original

2)      all lines being of equal length[13]

3)      all lines occurring as couplets

4)      rhyme being used

5)      a constant rhyme being used on even lines, plus optionally the first line

6)      each line exhibiting a caesura, or two in the case of qilü

7)      the lüshi exhibiting parallelism in the second and third couplets

Only the ‘true poems’ were analysed here, so in particular the literal translations (from various people) and the pure prose translations (from David Hawkes) were not considered.  Further, some of the poems found were not really translations, but were more just “based upon” a Du Fu poem.  Such poems were not considered either.  Most translations scored at least two points for matching the number of lines and being composed of couplets.  However the average translation scored fewer than three points in total, which very much supports the contention that the preservation of form during translation is a rare occurrence.  The one ‘most conforming’ translation is Stephen Owen’s “The View in Spring”, reproduced here:

A kingdom smashed, its hills and rivers still here,

spring in the city, plants and trees grow deep.

Moved by the moment, flowers splash with tears,

alarmed at parting, birds startle the heart.

War’s beacon fires have gone on three months,

letters from home are worth thousands in gold.

Fingers run through white hair until it thins,

cap-pins will almost no longer hold.

This scored six out of the possible seven, the one lack being the constant rhyme on even lines.  This particular example serves to illustrate some of the details of how the translations were scored and in particular how ‘generous’ the scoring was, which reinforces the rarity of conformance.  This translation qualified for the caesura point because although the last couplet does not have as well-defined a caesura as the others, it is possible to ‘force’ a caesura after the third word of each of these lines.  Secondly, the parallelism is not strictly word-for-word, but there is a grammatical structural equivalence exhibited by both middle couplets.  For example in the second couplet “the moment” can be considered to be grammatically equivalent to “parting”.  Finally, the rhyme point is awarded, even though rhyme is only exhibited in the second half of the translation, because the even-lines rhyme structure does match the original.

Another issue to be addressed is whether or not those translations that do show some of the qualities of conforming to the regulations of jintishi do so at the expense of other non-formal qualities.  To check this, all the translations were scored, rather subjectively, against the three ‘Yan Fu criteria’ of fidelity, comprehensibility and elegance, as mentioned in the introduction.  Each translation was assigned a score from 0 to 3.  The translations were then gathered into two groups, according to whether their jintishi score was above or below the overall average, which was 2.9.  Interestingly, the group with the low jintishi scores also on average scored lower ‘Yan Fu scores’.  The difference was not great, with the ‘less conforming’ group satisfying an average of 1.5 Yan Fu criteria, and the ‘more conforming’ group averaging 1.7.   The important point is that the more conforming group of translations did not on average score lower against the Yan Fu criteria, and thus in general, the conformance to jintishi regulations, as little as it is for the translations analysed, is not achieved at the expense of other criteria.

5. Some New Translations

The following translations are an attempt to provide an insight into what might be possible if more emphasis is given to preserving form than to preserving the literal meaning of the poem.  There is no claim that any of these translations is any ‘better’, however that is defined, than any other translation, but it is hoped that some merit may be found in this approach.  In each case some comments are provided.

5.1. wàng yuè

Gazing Upon the Sacred Mount

So what about              Taishan?

From all round              ever blue.

Creation of                   the Lord,

It splits the day             in two.

Clouds ease                  a troubled mind,

Far birds                      a strain to view.

Each must one day        ascend,

Look down on  all,        to do!

This translation preserves the even-length lines (six syllables each), and provides a caesura.  Although the caesura is not at a constant position, it is constant within couplets, and only one couplet deviates, so it would be awarded this point in the ‘generous’ scoring system.  It also provides the standard even-line fixed rhyme.  It does not preserve the parallelism in the third couplet, but parallelism is not necessary for gushi.  Thus in the scoring system described above it would score the full six possible points.

In the second line, blue is used instead of the more common green, since from a distance a mountain does look blue.  Either word is a valid translation of the original qīng, which can mean green, blue or black.  In the fourth line, hūn xiǎo is commonly translated as “dusk and dawn”, which doesn’t quite fit, since dusk and dawn are different times of day, rather than different amounts of light. The character hūn can also mean “dark”, thus a more appropriate literal translation is “dark and dawn” (or “night and day”) which is what is used here.

5.2. yuè yè

Night Moon (literal)

This night Fu zhou moon

Chamber in only alone watch

Distant pity/love little son daughter

Not understand remember Chang an

Fragrant mist cloudy hair-bun damp

Bright splendour jade arm cold/needy

What time lean/rely empty curtain

Two shine tear mark dry

This literal translation matches most other literal translations encountered with one exception.  In the seventh line the third character () is generally rendered as “lean on”, whereas an alternative valid meaning is “rely on”, which changes the meaning of the line from one of a hoped for physical activity, to a reinforcement of that expression of hope, which is in keeping with the romantic nature of the poem.

Tonight’s Moon

This moon        at home tonight

My wife            must watch alone.

I grieve for        my young ones

Who know not where I’ve gone.

Damp mist,       hair fragrance lifts;

Cool moon,      jade arm falls on.

When will         that bright-lit pane

Shine on           us both as one?

This translation would score a ‘perfect seven’ in the jintishi scoring system described above.  Note that although the second couplet does not exhibit the required parallelism for jintishi, the original poem does not exhibit parallelism in this couplet either, and so this ‘lack’ in the translation provides a closer match to the original.  Further the caesuras are not quite regularly positioned, but they are regular within each couplet.  In the fifth line a common translation has a “fragrant mist” making the “hair-bun damp” which reflects the word order of the literal translation.  Another interpretation is that the mist has become fragrant by lifting its fragrance from the hair bun which it has dampened.  This latter interpretation is reflected in the above translation.

5.3. chūn wàng

Spring Outlook (I)

Nation fallen, yet nature’s alive,

The city; spring trees and grasses thrive.

For these sad times the flowers they weep,

Being apart, birds stir me deep.

The war flames they’ll span three months soon;

Home news is worth a small fortune.

My white hair it’s torn out in vain,

Soon not to hold even a pin.

This first translation was an early attempt to provide a translation with rhyme in order to be ‘more accessible’ to a non-critical audience.  This would score four out of a possible seven points, since the rhyme is not fixed on even lines, there is no regularly located caesura in each line and couplets two and three do not adequately reflect the parallelism of the original.  The evenness of the length of the lines is debateable, depending very much on just how the stresses are voiced, but in terms of the scoring scheme applied to all other translations considered, this point would be awarded. 

Spring Outlook (II)

Nation fallen,                yet nature’s alive,

City in spring;               grass and trees bloom.

For these sad times,      the flowers they weep,

Being apart,                  birds deepen my gloom.

These war flames ere    will span three months;

Home news is worth     a small fortune.

My white hair   is          torn out in vain,

’Twill hold not e’en       a hairpin soon.

This translation satisfies the rhyme and caesura requirements missing from the previous translation, but still fails on the parallelism requirement, so would score six out of seven points.  Note that in line five the word “soon” is replaced by the (somewhat obscure) “ere”.  This is to avoid the use of a ‘rhyme word’ anywhere except in a rhyme position, thereby taking note of the long list of “things to be avoided” in jintishi (see Mair & Mei p.462-463).


5.4. yuè yè yì shě dì

Thinking of My Brothers on a Moonlit Night

Watchtower drums       interrupt travel,

Autumn Frontier -         a lone goose’s call.

From tonight on            the frosts will settle,

This moon’s as bright    as it is back home.

My brothers, they’re     scattered all over;

No home to ask            of their fate at all.

Letters are slow            to get wherever,

Besides, the war           is not yet done.

This translation does not satisfy the constant rhyme or parallelism requirement but does satisfy the other requirements, and so would score five out of seven points. 

5.5. tiān mò huái lǐ bái

Thinking of Li Bai at the Tip of the Sky

Cool winds freshen       at the tip of the sky,

Good friend just what   are you now thinking?

Wild goose, when will   it bring me your news?

Rivers and lakes;          by fall rains swelling.

Literature shuns             achievement of fame

Demons rejoice             to see men passing.

You ought to speak       with the wrong’ed soul,

Toss the Miluo               lines for his keeping.

This translation satisfies the length, couplet, caesura and rhyme requirements for lushi, and would score six out of seven points, failing only on the parallelism requirement.  Although the rhyme is not ‘perfect rhyme’ it does provide some aural ‘satisfaction’.

5.6. ké zhì

An Unexpected Guest

All around my house                 do the spring waters flow;

Nothing do I see                       but a flock of gulls daily.

Blossomed path has yet            for guests needed sweeping;

Wicker gate opens                    for you - first time only.

Far from market food               offers little flavour;

This poor household wine         is quite old and cloudy.

If you’d like to sit                      and drink with my neighbour,

I’ll call through the fence -         we’ll down it completely.

This translation comes close to satisfying all the requirements of jintishi, to the extent that the English language allows.

5.7. chūn yè xǐ yǔ

Spring Nights Welcome Rain

Good rain         knows its best time;

Come spring     and it is there.

On breeze        with night-time stealth;

Soaks all          with quiet care.

Walkways        are dark with cloud;

Bark light          shines all the more.

Dawn breaks    all red and wet;

Blooms fill        the city fair.


This is another first attempt that can probably be improved upon, but comes close to all requirements except the parallelism requirement.

5.8. wén guān jūn shōu hé nán hé běi

News that the Imperial Army has Recaptured North and South of the River


Beyond the pass – a fresh dispatch:                   the North has been retrieved,

When I first hear, from happy tears                   my clothes become all wet.

I turn to see my wife and kids;                           distress where has it gone?

Gathering up poems and books                         delight wants no curb yet.

Though heat of day, singing a song,                   in wine I must indulge;

In green of spring, joining my friends                  off home we gladly set.

From Ba Gorge here, leaving right now,            through Wu Gorge do we go,

Then down on through old Xiangyang town       to Luoyang home we get.


This translation satisfies the constraints on the number of lines, the couplets, the fixed length lines, the caesura (both minor and major), the rhyme and the parallelism of the middle couplets.  Thus it would score a ‘perfect seven’ in the analysis above.

5.9. lǚ yè shū huái

A Night Traveller’s Thoughts


Soft grassy bank           in the night breeze wafts,

Alone tonight,               this tall mast standing.

Stars fall on down         to the flat wide fields;

Moon rises from           the great course flowing.

Fame on its own -         what gets words noticed;

Old, sick servants         should be retiring.

Adrift in life -                just what am I now?

But a seagull -               purgatory; waiting.


This translation again comes close to satisfying all except the parallelism requirements of jintishi.

5.10. dēng gāo

Climbing High


Keen the wind, high the clouds,             apes call mournfully;

Pure the isles, white the sand,                birds are wheeling home.

Boundless the trees shed leaves -          rustling down to earth;

Endless the Yangtze flows -                  rolling on and on.

Journeys long, autumns sad,      always travelling;

Every year, many ills,                climbing this alone.

Hardships and regrets   have     frosted up my hair;

Wretched now I have just         given up the wine.


This translation satisfies the line count, line length, couplet, caesura (both major and minor) and parallelism requirements of jintishi.  The rhyme, however, is not quite strong enough to score both rhyme points.

5.11. jiāng nán féng lǐ guī nián

Meeting Li Guinian in the South


At Prince Qi's palace    I often saw you;

Before Cui Jiu's hall      I sometimes heard you.

Southern scenery -        is truly special;

As we meet again         here in life's autumn.


This translation satisfies the line count and line length requirements of jueju, and almost satisfies the caesura requirements – the minor caesura is not well defined.  These are the only requirements for this form of poetry, so this translation comes close to satisfying them.








6. Conclusions

Translation is a difficult process requiring many varied skills.  The translation of poetry requires even more skills to produce an acceptable product.  Taking this progression to the limit, the translation of Chinese Regulated Poetry (jintishi) offers the greatest challenge.

While most past translators of jintishi have concentrated on providing the most accurate representation of the source language poem, with all its subtleties of contextual meaning and frequent allusions, this emphasis has generally resulted in a prose or blank verse product, often also accompanied by explanatory notes.

Now while the full understanding of a poem can be curtailed by a lack of knowledge of all the context and allusion, in the case of jintishi this would be a relatively small loss because of the intrinsic physical beauty of the poem in both sight and sound.  For this reason it is appropriate to at least attempt to preserve some of these physical features of such poetry.

The results of analysing several hundred existing translations have shown that on average a translation which does exhibit some of the characteristics of the form of the original poem can do this without great loss of other qualities.  This implies that if it is acceptable for a translation to compromise some of the non-formal characteristics in order to preserve more of the physical beauty of regulated poetry then there is the potential for this to be developed much further.  Perhaps such a compromise is still not required for some more of the physical structure to be incorporated into a translation.

The process of translating literature is never complete and we can hope that in our lifetimes we can see the full glory of Tang dynasty poetry revealed to the West.


Alley, Rewi Peace Through the Ages: Translations from the Poets of China (Peking: R. Alley, 1954)

Alley, Rewi The People Sing: More Translations of Poems and Songs of the People of China (Peking: R. Alley, 1958)

Alley, Rewi Tu Fu: Selected Poems (Peking: Foreign Languages Press, 1964)

Ayscough, Florence Tu Fu: the Autobiography of a Chinese Poet, A.D. 712-770 (2 Volumes) (London: Cape, 1929, 1934)

Barnstone, Tony “The Poem Behind the Poem: Literary Translation as American Poetry” Manoa 11.2 (1999) 66-75. 9 Mar. 2007 <>

Barnstone, Tony & Chou Ping “Du Fu”, The Drunken Boat 3.IV (2003) <>

Bortolotti-Van Loon, Paola “Towards a Phenomenology of Translation Theory” Norwich Papers, 6 (1998): 101-109. 9 Jan. 2007 <>

Chan Sin-wai “Form and Spirit in Poetry Translation” The Humanities Bulletin 3 (1994): 100-111 21 Dec. 2006, <>

Chang, Edward C. “Understanding Jintishi or Chinese Regulated Verse” 14 Aug. 2007. Washington Chinese Poetry Society. 18 Aug. 2007 <>

Chang, Edward C. “What is Jintishi?” 24 Dec. 2006. Washington Chinese Poetry Society. 18 Aug. 2007 <>

Cooper, Arthur R. V. Li Po and Tu Fu (Harmondsworth, Eng.: Penguin Books, 1973)

Davis, A. R. Tu Fu (New York: Twayne Publishers, 1971)

Downer, G. B. & A. C. Graham “Tone Patterns in Chinese Poetry” Bulletin of the School of Oriental and African Studies, University of London, 26.1 (1963): 145-148. JSTOR 29 Jul. 2007 <>

Elsworth, Bente “Babel Revisited” Norwich Papers, 6 (1998): 3-25. 9 Jan. 2007 <>

Fang, Grace “Mirrors in the Mind: Chinoiserie in Ezra Pound's Translations of Chinese Poetry” Norwich Papers, 6 (1998): 85-100. 9 Jan. 2007 <>

Feng Zhi 冯至, ed. Du Fu Shi Xuan 杜甫诗选 [Selected Poetry of Du Fu] (Hong Kong: Daguang chubanshe, 1978)

Frankel, Hans H. “Review of Liu, James J. Y. The Art of Chinese PoetryHarvard Journal of Asiatic Studies, 24 (1962-63): 260-270. JSTOR 18 Jun. 2007 <>

Frankel, Hans H. The Flowering Plum and the Palace Lady: Interpretations of Chinese Poetry (New Haven: Yale University Press, 1976)

Fryer, John “Chinese Poetry” Transactions and Proceedings of the American Philological Association, 33 (1902): xcii-xciv. JSTOR 24 Apr. 2007 <>

Hamill, Sam “Sustenance: A Life in Translation” Manoa 11.2 (1999): 81-89 Project MUSE, 17 May 2007. <>

Hart, Henry H. The Charcoal Burner, and Other Poems; Original Translations from the Poetry of the Chinese (Norman: University of Oklahoma Press, 1974)

Hawkes, David A Little Primer of Tu Fu (Oxford: Clarendon Press, 1967)

Holcombe, C. John “Du Fu's Beautiful Ladies” 22 Aug. 2007. 22 Aug. 2007 <>

Hung, William Tu Fu: China’s Greatest Poet (New York: Harvard University Press, 1952)

Index” Quan Tang Shi [Complete Tang Poetry] 25 Aug. 2005. Xiao Yao Literary Arts Network. 9 Mar. 2007 <>

Jackson, Richard “From Translation to Imitation” 8 Oct. 2005. The University of Tennessee at Chattanooga English Department. 7 May 2007 <>

Liu, James J. Y. The Art of Chinese Poetry (Chicago: University of Chicago Press, 1962)

Liu, James J. Y. Essentials of Chinese Literary Art (North Scituate, Mass.: Duxbury Press, 1979)

Liu, Shih Shun One Hundred and One Chinese Poems (Hong Kong: Hong Kong University Press, 1967)

Liu, Wu-chi & Irving Yucheng Lo, eds. Sunflower Splendor: Three Thousand Years of Chinese Poetry (Bloomington: Indiana University Press, 1975)

Lovelock, Yann “Review of Cooper, Arthur (trans.), Li Po & Tu Fu” 12 Aug. 2007 <>

Mair, Victor H., ed. The Columbia Anthology of Traditional Chinese Literature (New York: Columbia University Press, 1994)

Mair, Victor H., ed. The Columbia History of Chinese Literature (New York: Columbia University Press, 2001)

Mair, Victor H. & Tsu-Lin Mei “The Sanskrit Origins of Recent Style Prosody” Harvard Journal of Asiatic Studies, 51.2 (1991): 375-470. JSTOR 19 Jun. 2007 <>

Mair, Victor H. & Tsu-Lin Mei “The Sanskrit Origins of Recent Style Prosody” in Luk, Bernard Hung-Kay & Barry D. Steben eds., Contacts between Cultures: East Asia: Literature and Humanities Volume 3 (Lewiston, NY: The Edwin Mellen Press, 1993). 6 Jan. 2006 <>

Minford, John & Joseph S. M. Lau, eds. Classical Chinese Literature: An Anthology of Translations. Vol. 1, From Antiquity to the Tang Dynasty (Chichester, NY: Columbia University Press, 2000)

Owen, Stephen The Great Age of Chinese Poetry: the High T’ang (New Haven: Yale University Press, 1981)

Owen, Stephen, ed. An Anthology of Chinese Literature: Beginnings to 1911 (New York: W.W. Norton, 1996)

Palfrey, Thomas R. “Literary Translation” The Modern Language Journal, 17.6 (1933): 410-418. JSTOR 8 Aug. 2007 <>

“Poetry” Wikipedia 18 Aug. 2007. Wikimedia Foundation, Inc. 18 Aug. 2007 <>

Rees, B. R. “Some Thoughts on Translation” Greece & Rome, 2nd Ser., 21.2 (1974): 111-127. JSTOR 23 May 2007 <>

Rexroth, Kenneth One Hundred Poems from the Chinese (New York: New Directions, 1965)

Rexroth, Kenneth “Tu Fu” in Weinberger, Eliot, ed. The New Directions Anthology of Classical Chinese Poetry. Bureau of Public Secrets, 17 May 2007 <,%20Poems>

Scott, Clive “Introduction” Norwich Papers, 6 (1998): 101-109. 9 Jan. 2007 <>

Stimson, Hugh M. T'ang Poetic Vocabulary (New Haven: Far Eastern Publications, Yale University, 1976)

von Zach, Erwin Tu Fu's Gedichte (2 volumes) (Cambridge: Harvard University Press, 1952)

Waley, Arthur Chinese Poems (London: Allen & Unwin, 1946)

Watson, Burton The Selected Poems of Du Fu (New York: Columbia University Press, 2002), Questia <>

Weinberger, Eliot Nineteen Ways of Looking at Wang Wei: How a Chinese Poem is Translated (Mt. Kisco, NY: Moyer Bell, 1987)

Weinberger, Eliot, ed. The New Directions Anthology of Classical Chinese Poetry (New York: New Directions Pub. Corp., 2003)

Wu, John C. H. The Four Seasons of T’ang Poetry (Rutland, Vt.: C. E. Tuttle Co. 1972)

Yan Fu, "General Remarks on Translation", trans. C.Y. Hsu, Renditions 1 (1973): 4-6

Yip, Wai-lim, ed. Chinese Poetry: Major Modes and Genres (Berkeley: University of California Press, 1976)

Yu, Pauline, Peter Bol, Stephen Owen & Willard Peterson, eds. Ways with Words: Writing about Reading Texts from Early China (Berkeley: University of California Press, 2000)

Zhong Weihe “Translation in ChinaTranslation Journal 7.2 (2003). 18 Aug. 2007 <>

Appendix A: Translation Sources

See for a list of the sources of the translations gathered for this project.

Appendix B: The Translations

See for the complete sets of accumulated translations of the poems discussed above.  For each poem, the original traditional Chinese text is provided, along with pinyin representation, the original Tang pronunciation of the last character of each line (to determine the rhyme), and the original tone of each character, where the level tone is represented by L and the three deflected tones, namely rising, departing and entering, are represented by r, d, and e respectively. This is then followed by all the translations discussed in this essay. (Corrections, particularly the names of those translators named as anonymous, and further contributions are welcome to

[1] The field of technical translation provides somewhat less flexibility in the finished product.

[2] Of course part of the process of deriving some enjoyment or entertainment from a poem may involve some kind of analysis, but this – perhaps unconscious – analysis is to be distinguished from analysis for its own sake.

[3] Counts vary between approximately 1150 and 1450, depending on whether individual poems or just separate titles have been counted – one title may have as many as twenty individual poems associated with it.

[4] Erwin von Zach has produced a translation of the complete works of Du Fu into German.

[5] Note that there is an implied rule here that all rhymes occur in the level tone.

[6] Mair and Mei conduct their discussion purely in terms of the five-character form, but the first two characters of the seven-character form are invariant over all other descriptions presented here.

[7] The assumption here is that all rhyme words have the same tone.

[8] The selected poems are listed in generally agreed order of composition.

[9] using the Downer and Graham rules

[10] Fryer is not included here since his description is incomplete.

[11] The last of the ten poems in the table is a jueju.

[12] though it can exhibit stress variations

[13] In this context, length is measured in terms of stressed syllables, which is not quite objective, since the location of spoken stress is not necessarily unambiguous.  For the purposes of this analysis a point was awarded for “even length lines” if all lines were approximately of even length.