The Holy Bible, one of the most carefully translated works, the Koran, the Nobel Prize-winning novel Soul Mountain, and other bilingual or multilingual texts as well as poetry are sampled for analysis for the identification of the UT.

Samples are randomly selected, ranging from complete texts to single sentences. In most cases, randomly-selected pages or passages are examined. The sample selected for analysis is Page/Chapter 30 when possible, or the first complete sentences of the randomly selected pages, or text chosen through other random methods. Whatever is selected is intended to represent the translation without bias. Due to limited space, each sample discussed in this paper is kept as brief as possible. The twenty-one tables, except Tables 5 and 6, illustrate only one feature of the TT, namely, one or two types of punctuation marks in most cases. It is not suggested that punctuation marks alone may accurately reflect the quality of a TT. It is reminded that the punctuation marks are a component of acceptable translations, and that the paramount goal of this study is to identify the UT. Punctuation marks are standard features shared by the great majority of the languages in the world. For example, one full-stop may equal one sentence in most cases, though a small number of languages (e.g. Thai) do not indicate the end of the sentence with a period. In our study, wherever the punctuation marks are presented, they are to be interpreted accordingly.

As of April 2006, the UDHR which was originally written in the six official languages had been translated into 326 languages, making it available in 332 (United Nations, 2006). A sample of three articles (10, 11 and 12) in nineteen languages including the six United Nations official languages is selected for examination. The number of sentences in each of the TTs is listed below.

The analysis indicates that the UDHR has been translated sentence by sentence, article by article, with the original format, sentence order, and meaning closely followed. None of the languages sampled have changed, reduced, or increased the number of the original articles. All of them show the same articles translated into the same number of sentences, except Solomons Pidgin (Pijin) that condensed the ST six into four in the TT, which, nevertheless, reveals no alteration of meaning.

Except in rare cases where the regional Bible societies cannot or do not have the authority to edit the text, as in the case of the Chinese TSPM and CCC that have printed a bilingual version of the Chinese Union Version with New Punctuation without adjusting the discrepancies of verses that were translated from different sources (Greek original versus Hebrew original), comparative analysis indicates that all the following biblical texts have been translated sentence by sentence within the context of the Bible: (1) The New Testament of Our Lord and Saviour Jesus Christ (1899) (in Greek and English) – 83 pages of 640 selected for analysis, (2) The Interlinear Greek-English New Testament (1960) – 64 pages of 1027 sampled, (3) The New Greek-English Interlinear New Testament (1990) – 34 pages of 913 selected, (4) The Septuagint Version of the Old Testament and Apocrypha with an English Translation; and with Various Readings and Critical Notes – 69 pages sampled out of 1378 analyzed, (5) The Parallel New Testament in Greek, Chinese [four versions], and English (PNT) – entire text of 749 pages examined, (6) Holy Bible: King James Version in English and Chinese – entire text of 1891 pages compared, and (7) Holy Bible: New Revised Standard Version (in simplified Chinese and English) – complete text of 1993 pages analyzed. Following are the results of a sample taken from the Parallel New Testament (United Bible Societies, 1997: 30-31) with “TT2: Chinese V. 1b” from another version in English and simplified Chinese (National TSPM and CCC, 1997) added for comparative purposes.

TT1 and TT2 are the same version of Chinese translation except that the former is in fanti (traditional Chinese characters) while the latter is in jianti (simplified Chinese characters). As indicated in the table, when the jianti version was converted from the original fanti edition, the number of sentences changed from 22 to 24 while the semicolons, from 14 to 9, but the examination of the two pages listed above shows no change in meaning or in words. Analysis indicates that translators often convert into sentences independent clauses with a subject-verb unit.

The mean of the sentences in the five versions of Chinese translation and the English and that of independent clauses of the six translations are tabulated below.

The source text of the sample page contains 19 end marks and 8 semicolons, totaling 27, while the average of the six translations, 22 sentences which include 13 clauses with semicolons, the nature of which implies that the TT could be interpreted as 35 sentences or independent clauses. Though the mean total, accepted as such, shows a difference of eight, all the six translations consistently adhere to the order of the ST sentence by sentence and in most cases clause by clause, within the sentences. The ST meaning has by and large been translated.

An analysis of Pickthall’s English translation (Eliyasee and Pickthall 2001) in the Arabic-English work of 605 pages and Ali’s The Glorious Qur’ân Translation of 423 pages (1999), Ali’s other version The Meaning of the Glorious Qur’ân Translation [with new punctuation] (2002), and Nadvi’s French translation Le Saint Coran (2001) demonstrates that the translations were rendered sentence by sentence, or rather, verse by verse. Verse 30 of “Surah 30” (Ali 1999: 266; Ali 2002: 283) is sampled for comparison with the ST as well as with Eliyasee and Pickthall’s English translation (2001: 401) and with Nadvi’s French translation (2001: 293). The following table shows the TT arrangements.

The analysis shows that all the translations transfer the ST meaning as interpreted by the translators, following the ST syntax closely. The difference between Ali’s 1999 version and 2002 version deserves further mention. As indicated in the table above, Ali follows the ST strictly in the 1999 translation, but it seems much harder to read and understand. His 2002 edition is re-punctuated, appearing to significantly improve readability. The result is that he uses three sentences instead of one but reduces the use of colons from three to one. His new edition also shows a change in the last clause, namely, “but most among mankind do not understand” has been changed to “but most among mankind understand not.” Pickthall’s translation reveals some difference in the understanding of the original text, but regardless of the differences, including those in the French translation, all the TTs have been translated following the ST syntax closely. The UT is the sentence.

Confucius Publishing has released the translation of The Lun Yu [Analects, 499 sayings] by Confucius in twenty-three languages, two of which are fanti Chinese and jianti Chinese, as detailed by the publisher. The ST in ancient Chinese has been rendered either phrase by phrase or sentence by sentence, but the UT is clearly the sentence. An example of the English translation is provided below (Confucius Publishing, 2006, Chapter 7, Verse 6):

Each translation comes with the original Chinese on top of it. In the example above, every three Chinese characters are translated into a verbal phrase: “Zhi yu dao” is translated into “Aspire to the way,” “ju yu de,” into “align with virtue,” “yi yu ren,” into “abide by benevolence,” and “you yu yi,” into “immerse yourself in the arts.” The necessary conjunction “and” is added, and so is the note. These four verbal phrases make up the saying. The unit of this translation is the saying, or grammatically the sentence. The translations of the other languages have been completed in the same fashion. Full texts are available for comparison at Confucius Publishing’s website: www.confucius.org.

The CCRF (Department of Justice, 1982) is one of the numerous government documents that have been and are still being published in Canada by the federal government and various provincial governments. On an annual basis, the Canadian Language Quality Measurement System alone assesses the quality of 300 million words of translation (cf. Williams, 2004: 3), ranging from law to reports, anything the governments need to publish. The authors’ contrastive analysis indicates that more than 99% of what they have read has been translated sentence by/for sentence. It is usually hard to tell whether the Canadian bilingual documents are translated from French to English, or vice versa. The following is an example, taken from Part 1 of the CCRF.

Note that the leading clause in the French version has a colon that is not found in the English version. Such are the differences one may find in Canadian bilingual documents. Nevertheless, the Canadian government’s official documents nearly always show the following features:

• ST and TT carry the same meaning with the same linguistic power and effect as the pairing language;

• Formal language is employed, though popular reading materials may use casual language;

• The two texts appear parallel in

  • headers, cardinal numbers, most punctuation marks, etc.,

  • paragraphing,

  • number of sentences,

  • syntax comparability.

All these may be found in the example above.

The English translation Patent Law of the People’s Republic of China indicates the same approach to translation as that of the Canadian government’s documents. A review of Article 30 (Government of China: 1992: 14-15, 51) reveals the following:

As the table indicates, the English translation has been rendered sentence for sentence following the ST syntax closely. The article contains one compound sentence with two subordinate clauses, each of which has its own subordinate clauses. A rather complex sentence structure has been used (though the words are relatively simple) in order to render the article into one English sentence to match the ST. The comparison of this law, among others, indicates that Chinese legal documents are translated into English sentence by sentence.

Soul Mountain (Gao, 2000b), translated by Mabel Lee from Gao Xingjian’s Chinese Nobel Prize-winning novel Ling Shan (Gao, 2000a), demonstrates that the 506-page TT was completed sentence by sentence. An ST sentence may have been rendered into two or more English sentences. By the same token, more than one ST sentence may have been combined and translated into one English sentence. In other words, a component, often a clause, may be translated into an English sentence while a ST sentence may be rendered into a clause. A comparison of the use of periods, semicolons, commas and dashes of the ST and TT of the first page of Chapter 30 (Gao 2000b: 167-168) provides a glimpse of the features of TT sentence structure:

As shown in the table, the number of TT periods and commas differs significantly from that of the ST. What should be noted is that the ST usage of the comma may have contributed much to the difference. While a comma is seldom used without a conjunction to separate two coordinate clauses in English, it is grammatically correct to use the Chinese comma to separate not only subordinate clauses in a complex sentence, but also coordinate clauses within compound sentences (cf. Huang 1986: 49; Su 1999: 54). As commonly found, many Chinese commas are used in the same way as English semicolons. Six of the ST commas seem to have been interpreted as periods, which could be one of the reasons why the translation has more sentences than the Chinese text. In two instances, for fluency or other reasons, two ST sentences were combined with a third one. An addition of five and a reduction of two result in thirteen sentences in the TT. Nevertheless, Lee follows the ST closely. Except for sentence combining, splitting, and the reverse order of words, word groups, and clauses, she follows the ST sentence by sentence. Hardly can one find an instance of sentence relocation. Lee translates by the sentence within the context of the novel, limiting the freedom of re-creativity mostly to the sentence. The contrastive analysis indicates little difference in the approaches taken by Lee and translators of non-literary texts.

One of the most-beloved literary pieces in China, Xi You Ji (Journey to the West), has been translated by Jenner, among others, into English (Wu 1990). The 933-page ST has been rendered into 1858 pages in English. Analysis of samples (Wu 1990: 1800) shows that the TT follows the ST closely, in the original order of narration or events described.

As revealed in the table above, the TT contains the same number of sentences as the ST. Analysis shows that while additional semicolons and fewer commas are used in the TT, they do not change the fundamental structure of the TT sentences. Except for the re-arrangement of the first two sentences of page 1800, the TT appears to have followed the ST in the following aspects: (a) sentences, (b) clauses, (c) phrases, (d) poetic lines, and (e) end marks.

It should be noted that this sample also contains a poem, which has also been translated sentence by sentence, or rather, line by line or couplet by couplet.

The short-short stories written by the world-renowned Pu Songling have been translated into English. Rose Quong, for one, translated forty of his works (P’u [Pu], 1946). Despite a lack of translation theory in the 1940s, Quong’s English translations were successful, and by today’s standard, they are still of high quality. The first page of “The Fox Maiden Lien Shiang” (P’u [Pu], 1946: 27) is sampled for analysis regarding her approach to the treatment of ST sentences. The following table, through an analysis of three punctuation marks, indicates the differences in the sentence structure of ST and TT.

The ST contains eight sentences, while the TT, twelve with three semicolons present as well as four colons preceding direct speech. Four of the commas in the ST seem to have been interpreted as semicolons by the translator. Apart from the formal differences between the ST and TT, the following are observed:

• A cultural point, ming and zi are combined into “name” only, losing the translation of the character’s name – Xiao. In ancient China, ming was the given name, while zi, often another name indicating the moral aspect or fine personality of the name bearer. There is no equivalent in English, and for literature, such a loss causes no practical harm, and perhaps may be better left out to improve style.

• The translator follows the ST closely, resulting in sentences that may contain as many as four clauses which she separates, except for one, into shorter coordinate clauses each containing two sub-clauses or less.

• All the TT sentences follow those of the ST one by one. Where sentences contain three or more verbs, the translator identifies the independent ones for new sentences or independent clauses. Nevertheless, syntactically and semantically ST sentences are rendered directly where possible.

Utopia available in a Latin-English version (More, 1965) has been translated into Chinese (Dai, 1982). The Chinese, believed to be a translation of the English, was done sentence by/for sentence, or from clause (independent), or CI, into sentence, or vice versa. This approach is clearly demonstrated by the 133-page Chinese translation when it is compared with the English text of the 202-page bilingual Latin-English original. Page 30 of the Chinese translation, including the seven characters on the next page, shows the following against the English:

The 175-page Ma Ke Bai (Zhu [tr.], [1944]1996: 30), a Chinese translation of William Shakespeare’s Macbeth, was rendered sentence by sentence, or CI by CI, or line by line, following the order of the ST acts and scenes. Page 30 of Ma Ke Bai is sampled for analysis:

All the TT sentences follow the ST order except the change of line order in the middle of the page, and except for his failure to translate effectively “nor keep peace between the effect and it” (Zhu [tr.], [1944]1996: 30). An explanatory approach is also detected like in the case of “unsex me here” – the Chinese translation of which means “take away the gentleness of the fair sex.” In general, Zhu translated verse by verse or sentence by sentence semantically.

A comparison of pages 9, 19, 99, and 199 (Montgomery, 1993) of the French translation of The Story Girl indicates that the TT follows the ST chapters, paragraphs, and sentences closely.

While the ST contains more sentences than the TT, the TT has not changed the order of the ST sentences. The sentence-by-sentence comparability of the ST and TT indicates that the UT is the sentence.

“Of Studies” by Francis Bacon has been translated into Chinese by Dong Xu and ten other translators (see Feng, 2002: 390-406). The one-paragraph essay consists of 19 sentences including thirty-one semicolons and two colons, but the number of sentences in the Chinese translations differs – from twelve to thirty-six (Feng, 2002: 406). Needless to say, the number of semicolons in the ST indicates that the translation could potentially be rendered into fifty independent clauses. The average of the TT, however, is twenty-one sentences with 787 Chinese characters. Compare the TT means with those of the ST:

As indicated in Table 14, the TT contains more characters than ST words. One explanation is that an English word often has to be translated into more than one Chinese character as in the case of Bacon’s two-word title, “Of Studies,” which was translated into three characters – “Lun du shu.” The larger number of sentences in the TT, on the other hand, may be a matter of style.

What the eleven translations have in common is that each has been translated sentence by sentence and that each is individually comprehensible. The eleven translators not only follow the ST sentence order, but also the order of the clauses within the sentences in most cases. The sentence appears to be the UT.

Contrastive analysis of a Chinese translation of Blake’s “The Tiger” (Jiang [tr.], 2006: 30), a Korean translation of Meyer’s “Pemberton Park” (Huh [tr.], 2005: 30), Xu’s English translation of Yang’s “Zong Yang Gong Wang Yue” (Xu [tr.], 1997: 30-31), two English translations of Cao’s “Hao Liao Ge,” entitled by Chinese readers (Cao, [1792]1992: 10) by Hawkes (Cao, 1973: 63-64 [TT1] ) and Yang and Yang (Cao and Gao, 1994: 21-22 [TT2]), among many others, reveals similar features:

Table 15 indicates no change in the number of stanzas and lines. Further, the authors have found that translators closely follow the order of the lines as well as subject-verb units wherever possible. Despite the different TT punctuation marks, the TT lexical content matches that of the ST. The semantic sentence is clearly the UT.

Two self-translated poems, “Thirst: A Lion,” and “That Word” by Zheng (1992: 196-199) demonstrate an approach to translation similar to those in the section above.

The use of the period and semicolon in the middle of “That Word” demonstrates that self-translating authors take a more flexible approach to translation than translators. A translator does not always add end marks to the TT unless they are present in the ST, but Zheng adds twelve in the self-translated text. Nonetheless, in principle, Zheng’s approach resembles that of the majority of translators – it is sentence-by/for-sentence, or line-by-line.

Gabrielle Roy Prize-winning author Lien Chao’s free verse directly written in Chinese and English resembles translation. A sample taken from her poetry (Chao, 2004: 28-33) shows the following:

Chao’s bilingual text shows more differences in syntax than the translated poems discussed previously (from Chinese to English). There are two instances where one line is split into two, and there are three cases where two lines are combined into one. There is an example of three lines reduced into two, while in two cases two lines are converted into three. The analysis suggests that translators seem to impose more restrictions on translating than writers on bilingual writing. Or there could be a possibility that authors care more about meaning than form. Despite the differences, Chao’s English text matches the Chinese in sentence order. The poet appears to maximize the equivalence of meaning in the two languages sentence by sentence. Indirectly, Chao confirms the sentence as the UT.

The responses to the first question of the international survey are shown below:

Note that one multiple response with three answers is excluded. As the table above indicates, 86% of the published translators translate sentence by sentence within context. It agrees with the analysis of the bilingual/multilingual texts presented previously. The results of the survey confirm the sentence as the UT. They are further supported by the results of a backup survey among thirty professional writers, twenty-one of whom were among the sixty-six who responded to “How do you translate?” Undeniably, 14% is still a significant percentage, even though a small minority. Further study of this group is needed to find out what their UT may be. Could the discrepancy be attributable to a lack of definition of UT in the survey (ST vs. TT)? Or could it have resulted from the respondent’s wrong judgement of his or her own UT in use? (The numbers in the next section show that up to 96% often translate sentence by sentence within context.) But one thing is clear: for the overwhelming majority of translators the sentence is the UT.

Question 2 of the translators’ survey further confirms that 87% of translators translate sentence by sentence within context more frequently than often, and that 96% often translate sentence by sentence, while 4% seldom or never translate sentence by sentence. Three invalid answers were excluded. What deserves attention is the participants’ interpretation of the answers, for example, “E. Often: 50% (or 40-59%).” Is the hypothetical number 50% considered a “yes” or “no” answer to “Do you translate sentence by sentence?” If “often” (50%) is a “yes” answer, there would be a 96% majority, which thus creates a discrepancy between the answers to the first and second questions. Otherwise, while the answers to Question 1 show that 86% of the translators translate sentence by sentence, Question 2 indicates that 87% translate sentence by sentence more than often (70% of the time), in which case, both answers confirm each other when the accuracy difference is factored in. Regardless of the possible gap, the answers to Question 2 further confirm the sentence as the UT for the overwhelming majority of translators. In percentage terms, the average translator renders 82% of his or her translation sentence by sentence within context.

It is worth noting that, excluding five invalid answers, 81% (21 of 26) of the published writers do 50% or more of their writing sentence by sentence, and that the average completes 63% of his or her writing sentence by sentence. The results of the survey among the students show that 90% complete 50% or more of their writing sentence by sentence, and that the average student writer completes 70% of his or her writing sentence by sentence. The two backup surveys further confirm that the sentence is the unit of writing for the majority of writers, which explains why translators make the sentence their UT. If writing is done sentence by sentence, it is only natural for translation to follow suit.

The results of the surveys also indicate that for the majority the sentence is the unit of translation evaluation and editing. Seventy-four percent of translators and translation editors indicate that they evaluate or edit translation sentence by sentence within context.

As Table 20 shows, 16% of the translators or editors evaluate or edit translation paragraph by paragraph, which corresponds to the second most important group. It is likely that many translators and editors may take a multiple approach in practice, but their answers clearly indicate that they know of the primary unit of evaluation and editing, which reconfirms the sentence as the UT.

In contrast, no clear majority of writers and editors use the sentence as the primary unit of evaluation and editing. While that unit may be the sentence, the paragraph plays an even more important role in this regard. Seventeen percent (5 of 30) take a multiple approach. If the multiple answers are included and each item is calculated based on the total number of participants, 47% indicate the sentence as the unit of evaluation and editing, and 53%, the paragraph.

The student-writers’ survey, on the other hand, shows that 70% (51 of 73, excluding multiple answers) use the sentence as the primary unit, a significantly higher number than that of the writers/editors group, probably due to shorter lengths of writing. These numbers suggest that, unlike the unit of writing, that of translation evaluation and editing and that of translating, either the sentence or the paragraph could be the writers’ unit of evaluation and editing, though paragraph connection and overall stylistic matters and other issues are not meant to be overlooked. As it may be interpreted, due to the likelihood that ST editing is conducted at different levels through a variety or combination of approaches, TT editing and evaluation at the sentence level may suffice to guarantee the essential quality of the ST. In other words, the ST editor may have completed much of the work regarding themes, logic, structural organization, etc., above the sentence level, which allows the translator to translate and evaluate the sentence within the textual context.