Chinese Translation of Literary Black Dialect and Translation Strategy Reconsidered:

The Case of Alice Walker's The Color Purple

Abstract

This paper aims to adopt corpus-based research to the translation of seven unique syntactic categories of African-American Vernacular English (AAVE) into Chinese. In this study, the parallel corpus consists of Celie's use of black dialect in The Color Purple and its three Chinese translations produced in Taiwan by Hui-quian Chang, Zu-wei Lan and Ji-qing Shi. To avoid subjective speculation, corpus-processing tools are used as an aid to spot linguistic patterns or creative renderings in their translations. It was found that these three translators' renderings of the seven syntactic features conform to the idiomatic expressions and syntactic structures of the target language. The readability of the three Chinese translations remains high, and such rendering inevitably results in fluent translation. Lawrence Venuti has warned translators that a fluent translation can produce detrimental effect on minimizing the foreignness in the source text. To remedy this problem, this study suggests that translators who choose not to represent the distinct syntactic features of AAVE should inform readers of the unique linguistic characteristics of AAVE in a preface or postscript, so readers have a chance to appreciate the writer's deliberate use of a particular language style.

Keywords: corpus-based approach, literary translation, syntactic patterns of black dialect, Alice Walker, The Color Purple

lice Walker, whose southern background has a profound influence on her writing and her permanent concern about the black woman's struggle, is a highly acclaimed African-American writer of novels, poems and essays. Her most well-known novel, The Color Purple, is about the life of a poor black woman named Celie and her struggle with her stepfather's rape, her husband's abuse, and her more than thirty-year quest for independence. Heeding her stepfather's warning about "telling nobody but God" because "it'd kill your mammy" (Walker, 1983, p. 1.) at the very start of the story, Celie writes letters to God to express her distress. Traumatized physically and psychologically, her predicament is conveyed by means of African-American Vernacular English (AAVE) in these letters. This deliberate use of AAVE is considered to be a rich representation of black folk's speech, but it has posed unusual challenges to translators who attempt to represent and preserve its unique linguistic qualities.

Walker's attempt to write in AAVE in this novel presents a knotty problems for translators. Not much attention has been paid to how translators tackle the problem of rendering the unique linguistic characteristics of AAVE in the translation into Chinese. Therefore, our study is intended as an investigation of how distinct syntactic features of AAVE in The Color Purple are rendered in these three Chinese translations within the framework of a corpus-based analysis. To avoid subjective speculation, corpus-processing analytical tools are also used as an aid to spot linguistic patterns or creative renderings in the different versions of translation. This paper consists of four sections. Section 1 briefly summarizes the characteristics of seven syntactic features of AAVE. Section 2 reviews scholarly works relating to the translation methods for rendering literary black dialect. Section 3 reports on the design of the English-Chinese Parallel Corpus of African-American Vernacular English (PCAAVE) for this study, search items used to examine the texts in the corpus, and two corpus-processing tools used in this study. Section 4 discusses the results of the research and reflects on the translation method employed by the translators. In conclusion, feasible solutions to the problem of achieving an equivalent linguistic effect due to the linguistic incompatibility between AAVE and Chinese are provided.

In The Color Purple, AAVE used by Walker to represent Celie's speech style is also known outside academic circles as Ebonics, black dialect, or Black English Vernacular. Its origin has been controversial because it mostly refers to the vernaculars of descendants of slaves in the United States of America. It is argued that it arose from the contact between speakers of West African languages and speakers of English. That is to say, AAVE derived from the institution of slavery in the South between the late 16^th century and mid-19^th century. Then, with the industrial revolution and the Civil War, the black dialect was brought from the southern plantations and ranches to the cities in the north and all over the country.

There are seven distinct syntactic features that distinguish AAVE from other dialects of English and American Standard English. As a matter of fact, the seven syntactic features listed below can all be found in The Color Purple. Firstly, AAVE has a number of ways of marking negation. The use of ain't is a general negative indicator in a simple sentence. It is used in place of "am not" in example (1a), "isn't" in example (1b) and "aren't" in example (1c) or even "didn't" in example (1d):

AAVE also has a special negative construction called "negative inversion" by linguists. In a negative-constructed sentence, an indefinite pronoun such as nobody or nothing can be inverted with the negative verb particle for the purpose of emphasis (see example 1e). Example (1e) also demonstrates another apparent characteristic of AAVE, the so-called "double negation."

When there are three or more negative forms in a sentence, "multiple negation" like that in example (1f) occurs. Both double negation and multiple negation are stigmatized by users of American standard English, because it is assumed that double negation leads to positive implication. In fact, if an AAVE sentence is negative, all negatable forms are negated.

Secondly, inflected forms, including genitive and the subject-verb agreement in simple present tense are often omitted. Example (2a) shows that there is no -s ending in the present-tense verb when the subject is the third person singular. Example (2b) shows the missing possessive form of non-pronoun, ending 's:

Thirdly, two kinds of wh-questions come into sight in AAVE: (1) without inversion of be or auxiliary verb (see example 3a and 3b) and (2) without be or auxiliary verb (see example 3c and 3d).

Fourthly, American Standard English uses a conjugated be verb, called "copula" in a number of different situations. It may occur as is or 's, are or 're, am or 'm, etc, whose occurrence is determined by the subject. In AAVE, it is found that copula is often omitted in the present tense (see example 4a). Moreover, appearance of be is frequently used to indicate events that occur habitually or repeatedly(see example 4b).

Fifthly, future events and those that have not yet occurred are marked by gon or gonna. Example (5a) is a sentence with gonna while example (5b) with gon.

Sixthly, the absence of have or has in a present-perfect sentence is often mistaken by users of American Standard English. With have or has omitted, an AAVE sentence might mislead users of American Standard English that the action occurred in the past. Example (6a) in AAVE actually means that she is still a teacher now. Nevertheless, for users of American standard English, the inference will be quite opposite: namely that she is now no longer a teacher. Moreover, AAVE sentences with done are designed to emphasize the perfective action (see example 6b).

Finally, simple present progressive sentences emerge without be in AAVE (see example 7a). In addition to using the verb with the ending -ing or -in to convey that an event is in progress, users of AAVE add steady in sentences like example (7b) to imply actions that occur consistently or persistently.

There may be many different versions of translations of works written in AAVE; however, how translators deal with the original author's attempt at writing in dialect remains an unsettled problem. As Hatim and Mason indicate in Register Membership in Literary Translating, "Translators may opt for the classical variety throughout (version 1), one of the vernaculars throughout (version 2) or one of the vernaculars for less formal speech and the classical for more formal speech (version 3)" (1997, p. 99). On the question of whether linguistic style of dialect in literature be well preserved by means of local language, Stog in Reflections on the Problem of Dialects' Translation, points out that the translator who wishes to render literary dialect into local linguistic forms will produce translation "with all the resulting problems of incongruity and misplacement" (2006, p. 84). She thus concludes there is no completely satisfactory way of translating the varieties of dialects mentioned by the author due to the difficulty of finding equivalent varieties that properly associate with a place or a social group in the target culture. Two studies, however, provide different insights into the subject of rendering of AAVE into a local dialect.

Gloria & Herman Wekker's study focuses on the feasibility of using Surinamese Dutch for the translation of AAVE in The Color Purple. In Coming in from the Cold: Linguistic and Socio-Cultural Aspects of the Translation of Black English Vernacular Literary Texts into Surinames Dutch, they cite ten linguistic characteristics of AAVE: (1) absence of copula and auxiliary be (2) absence of past tense marking (3) absence of third-person singular -s (4) absence of possessive -s' (5) absence of personal pronoun subject (6) absence of subject-auxiliary inversion in questions (7) absence of subordinator that (8) use of double negation (9) use of -s with persons other than 3^rd person singular (10) use of subject repetition. They criticize the earliest Dutch translation of The Color Purple produced by Irma van Dam in 1983 as inauthentic and is a product of her invention because no Dutch person would ever speak or write in that way. Since features of AAVE are present in the original, "the TL text must try to achieve a similar effect on the foreign reader as the SL text does on the native reader" (1991, p.221). They insist that "the main purpose of literary translation is to engender similar feelings and reactions in the reader," and it is obligatory for the translator to "consider non-standards in the target culture which are capable of transferring the specific socio-cultural characteristics of the non-standard used in the SL text" (1991, p.228). Their major contention is that in terms of linguistic and socio-cultural aspects, Surinamese Dutch serves as an adequate equivalent for the translation of AAVE into Dutch in terms of the similarity of linguistic and socio-cultural aspects of these two languages.

Ju-yu Lin's master thesis, Translation and Commentary of Mark Twain's The Adventures of Huckleberry Finn (2001), is the earliest research project in Taiwan that considers how to preserve the unique language style represented by Mark Twain. From her personal practice of translating the first thirteen chapters, Lin tend to embody Jim's dialect by means of Taiwanese and Taiwanese Mandarin, which also make use of non-standard pronunciation. Lin divides the linguistic characteristics of the AAVE in the novel into two parts: the nonstandard pronunciation and grammar. The former refers to nonstandard spelling and words with apostrophes occurring because of a glottal stop. The latter includes five grammatical features: (1) the discrepancy of the subject person and the verb form (2) the shortage of the auxiliary verb (3) the nonstandard verb tense (4) the nonstandard verb form (5) double negation. To render nonstandard spelling and words with apostrophes, she uses incorrect Chinese characters. She also incorporates popular expressions used by younger generation in Taiwan into her representation of Huck's talking style. Because AAVE is sometimes considered a nonstandard language by people who speak American Standard English, Lin thinks the source text can be rendered in Taiwanese and Taiwanese Mandarin.

So far there is still a lack of research on how the syntactic features of AAVE in The Color Purple are tackled in the translation into Chinese and what translation method is used by the translator. It is thus necessary to look in more depth into the Chinese translators' choices in dealing with the deliberate use of AAVE in The Color Purple. With the development of technology, corpus-processing tools enable the investigation to be more efficient, and corpora can be put together for any particular purpose and expanded when required. Corpora that consist of machine-readable texts have become data banks that allow researchers to look into various translated texts at the same time and conduct a more objective qualitative analysis, rather than merely subjective speculation.

3. The Construction of PCAAVE and Corpus-Processing Tools

A parallel corpus is one of several kinds of corpora, consisting of a set of texts in one language and their translations in another. That's why Stig Johansson (2004) called it a "translation corpus" rather than "parallel corpus." Nevertheless, the term "parallel corpus" is more widely used in the field of corpus-based translation studies. For the purpose of this study, the English-Chinese Parallel Corpus of African-American Vernacular English (PCAAVE) is designed as a bilingual (English and Chinese), parallel and uni-directional (from English into Chinese) corpus.¹ It consists of Celie's part as well as its corresponding translations into Chinese by Hui-qian Chang, Zu-wei Lan and Ji-qing Shi. As special emphasis is to be placed on the investigation of the renderings of syntactic features of AAVE into Chinese, the corpus constructed for the present research therefore contains only letters with Celie's signature and their translations.

With the corpus on hand but without any search criteria to carry out the examination of a large quantity of data, the PCAAVE will be of little use. The seven syntactic features of AAVE discussed in section two are used as search items to investigate each translation. That is, based on each syntactic feature, one or more search words are selected for further analysis and evaluation. The search words for certain syntactic features of AAVE are listed in Table 1:

Table 1 Search Criteria for Corpus Analysis

Due to the limitation of word search processes in corpus-processing tools, the following syntactic features may require manual examination: (1) omission of inflected forms (including third-person singular and genitive) (2) wh-questions without the inversion of the subject and be or auxiliary verb (3) wh-questions without be or auxiliary verb (4) omission of be in the present tense (5) present-perfect-tense sentences without has or have. Some sentences are listed in the Table 1 as examples for reference.

As the purpose of our empirical study is to identify, describe and analyze features of translated texts, researchers in corpus-based translation studies rely tremendously on the application of corpus software tools, such as WordSmith and ParaConc, to process a large amount of data in a time-consuming way. Designed by Mike Scott and published by Oxford University Press in 1996, WordSmith is a popular monolingual software package for lexical analysis. Version 4.0, the latest one, is used in this study. Being widely employed in corpus linguistics, WordSmith is capable of processing both single-byte Indo-European languages such as English and double-byte languages like Chinese, and provides basic linguistic information of the source text and the target text, including statistics for a text (text length in words, sentences, or paragraphs and type-token ratio), frequency list, word list in alphabetical or descending order, and basic collocation statistics.¹

ParaConc, a bilingual or multilingual concordancer, plays a vital role in contrastive analysis, especially in the parallel corpus for researchers' purpose of comparing the source text and its different versions of translation. Its first appearance was a Macintosh version in 1995, and Michael Barlow designed a Windows beta version the following year. The present study employs ParaConc 269, which is the latest build. In addition to the Indo-European languages, Font Support by ParaConc makes it possible to examine Chinese texts. ParaConc currently can handle four texts at most, which means that Walker's source text can be compared with three Chinese translations at the same time. The advantage of running four texts simultaneously is that it not only aids the purpose of judging the source text against any one of three translations, but it is also useful for evaluating the differences between the three translations.

PCAAVE is constructed to examine the translational features of AAVE manifested in three Chinese translations of The Color Purple. The statistical results of text length generated by WordSmith first show that both Chang's translation (52,709) and Lan's translation (59,869) are longer than the original (52,145). Shi's translation is the shortest due to the omission of several passages about the sexual relationship between two of the female protagonists, Celie and Shug. If the passages Shi has omitted are also deleted from Walker's original, there will be more words in Shi's translation (50,910) than in the source text (47,944).

To disclose how syntactic features of AAVE are rendered in the three Chinese translations, ParaConc is then employed to investigate each syntactic feature of AAVE in the parallel corpus by the search words listed in Table 1. In what follows, the syntactic features of negation, omission of inflected forms, wh-questions, copula, future events, present perfect tense and simple present progressive tense are discussed in detail, with examples gathered from the PCAVVE by ParaConc.

First of all, in rendering ain't in the source text, "不", "不是", "沒有" and "沒" are the four most common translated words or phrases used by the three Chinese translators. They account for 73% of Chang's sentences with negation, 70% for Lan, and 72% for Shi. Example 1 shows that Walker's original reveals a high-level of certainty. To make the tone more explicit, Chang turns the question into a declarative sentence with affirmative tone. Lan adds an extra question tag "不是嗎" to show assurance in addition to the similar construction of declarative sentence.

Shi is an exception to the rule that is found in Chang and Lan's translations. Most of Shi's sentences of general negation are carried out by literal translation. For instance, example 1 above is rendered literally without any change in syntactic structure. Shi has only one sentence that renders ain't non-literally (see example 2).

Negative inversion happens when an indefinite pronoun such as nobody or nothing can be inverted with the negative verb particle in a negative-constructed sentence. The purpose of negative inversion is emphasis. Searched in the PCAAVE, five sentences with negative inversion in the source text are found. When it comes to negative inversion, the three Chinese translators come up with similar rendering. Negative indicative words such as "沒", "沒有" and "不"(see example 3) are the most common device to reproduce negative inversion:

Yet, Lan's translation in boldface in example 4 is so particular that it deserves our attention. Lan makes good use of a typical four-character Chinese idiom "無憂無慮" to convey the original meaning. His expression is more target-user oriented, but the tone is more forceful than Chang and Shi's literal renderings.

Double or multiple negation is another well-known linguistic feature in AAVE. No matter how many negative parts constitute it, a sentence still remains negative. The three Chinese translators utilize similar translated phrases such as "不", "沒有" and "沒" to render this syntactic feature. The statistics shows that Chang translates 75% of these sentences in this way, Lan 71%, and Shi 69%. It is found that the three translators are inclined to clarify the meaning of the original rather than offer a literal translation that could be misunderstood. The reason for clarification is that an AAVE sentence with double or multiple negation rendered in literal Chinese might be easily interpreted as positive.

Double or multiple negation in AAVE actually confuses the three Chinese translators. Both Chang and Lan have eight sentences mistranslated, and Shi has three. The following examples 5 and 6 are two misinterpreted sentences of double and multiple negation. Lan's translation in example 5 renders double negation with a positive meaning. Because the second negative word, nothing, doesn't influence the negative effect of the whole sentence, it is the first negative word, don't, that produce a negative meaning. Moreover, if translators choose to do the literal translation in AAVE sentences with double or multiple negation, the possible translation of example 10 could be rendered as follows: 看起來不像沒有什麼. This kind of translation sounds awkward and might impede readers' comprehension.

Likewise, in example 6 Shi also makes a mistake in rendering a multiple negation. In fact, hadn't, instead of hardly and no, produces the negative meaning:

The mistranslations account for a fairly small percentage of all sentences with double or multiple negation constructions. It would perhaps be fairer to say that these mistakes were made due to carelessness during the translation or proofreading process, rather than a lack of understanding of double or multiple negations.

Secondly, possessive forms of non-pronoun ('s) are often omitted in AAVE. The statistics shows the high percentage of translators' understanding that the omission of the genitive is one of the syntactic features of AAVE. All three Chinese translators are inclined to comprehensively render the syntactic feature as "Noun＋的＋Noun" rather than simplify it as "Noun＋Noun". As in 120 sentences, Shi's renderings are all correct; both Chang and Lan have two mistranslations. Example 7 shows that Chang and Lan misunderstand the original and come up with translations without possession relationship between the adjacent two nouns due to their omission of possessive forms.

Owing to the linguistic difference between AAVE and Chinese, some syntactic features of AAVE cannot be manifested in the three Chinese translations. English is a branch of the Indo-European family of languages and needs to specify the person or tense by means of an overt inflection mechanism. Chinese, however, belongs to the Sino-Tibetan family of languages, and has no similar morphological resources. Unlike English, tense and aspect are not grammatical categories in Chinese. Neither does the form of the verb change in accordance with a time relationship. To deal with those unique syntactic features of AAVE, Chinese idiomatic expressions and adverbial phrases are chosen by the three Chinese translators. Users of Chinese are used to relying mostly on adverbial phrases or the context to express and comprehend the time relationship in the text. The following is a summary of Chinese renderings of these unique syntactic features of AAVE mentioned above.

Firstly, appearance of be in an AAVE sentence is frequently used to indicate events that occur habitually or repeatedly. In the PCAAVE, fourteen sentences with this syntactic feature are found, and Chinese translated phrases like "每次", "經常" or "時常" are used to specify the routine-like incident. In example 8, only Lan's Chinese rendering "每次"and "那水都開了"reveals to the readers that this is a habitual event, which occurs again and again. Chang's and Shi's renderings do not translate the word be:

If temporal adverbial is specified in the source text as shown in example 9, the three Chinese translators render the literal meaning of adverbial phrases to deliver the effect of repetition. Please see example 9:

Secondly, in order to explain something that has not yet occurred and is going to happen in the future, translated words or phrases such as "會", "要", "將會", "想", "打算" and "欲" are used to convey the meaning of AAVE sentences with gon (see example 10) or gonna (see example 11).

With the use of these phrases, readers understand that the event has not happened yet and will happen in the future. Chang's use of these translated phrases accounts for 43.7%, Lan's for 36%, and Shi uses them most often, in 77% of 57 instances.

Thirdly, a present-perfect-tense sentence without has or have in AAVE carries the same implication as that with has or have in American Standard English. In rendering sentences without has or have, translated words or phrases such as "了", "過", "一直", "已經" and "自從" are used by three Chinese translators. 70% of Chang's sentences are rendered in this way, while Lan has 72%, and Shi has 60%. Moreover, done is used to emphasize the completed nature of the action in AAVE. "了" and "已經" are favored translated words and phrases for the three Chinese translators to render done in their translations (see example 12):

Fourthly, verbs ending in -ing imply that the event is in progress, a syntactic feature shared by AAVE and American Standard English. The difference lies in the fact that there is a copula in the latter and only the verb ending in -ing appears in the former. The statistics show that nearly 90% of the sentences with verbs ending in -ing are not rendered by translated phrases like "正在" to indicate the progressive activity. Rather, translators choose not to render the progressive meaning embedded in the sentences with events that are happening right now.

Another indicator, steady, is used to mark actions that occur consistently or persistently (see example 13 and 14). The following two examples show steady are rendered with adverbial phrases in boldface:

In example 13, Chang's rendering "繼續" shows that the protagonist continues to do something. Readers of Lan's rendering "持之以恆" know the protagonist keeps the habit of doing something without giving up, and Shi uses 一直 to tell readers that the protagonist does something all the time.

However, Chang and Shi in example 14 do not pay attention to steady. Only Lan uses the Chinese adverbial phrase "一次又一次" to show the consistent occurrence of an event.

Finally, the non-inversion between the subject and be or auxiliary verb and the omission of be or auxiliary verb in wh-questions may not be accepted by users of American Standard English. See the following examples 15 and 16 for references on the non-inversion of the subject and auxiliary verb:

This syntactic feature does not hamper the Chinese translators' understanding of AAVE sentences because of the question mark at the end, as well as the various interrogative keywords at the beginning of the sentence. Both examples are rendered in the form of direct questions except Lan's translation in example 16. He turns the original questions into declarative sentences, making the tone in the original more explicit and emphatic:

The above analysis shows that the following syntactic features of AAVE are not preserved in the three Chinese translations because it is impossible to find any Chinese equivalent syntactic feature: (1) omission of copula in present tense (2) be as an habitual indicator (3) gon or gonna as a future-event marker (4) omitted has or have in the present perfect tense (5) done as an action-completed signal (6) omitted be in simple present progressive tense (7) steady as a consistent marker. A high percentage of these features are rendered using Chinese idiomatical expressions or adverbial phrases. A wide divergence of syntax between AAVE and Chinese does exist and causes difficulties to translators seeking to represent the linguistic characteristics of AAVE and to produce an equivalent linguistic effect in Chinese translation.

Concluding Remarks

The aforementioned analysis of the syntactic features rendered in the translation into Chinese shows that the translators tend to minimize the uniqueness of AAVE by using grammatical Chinese renderings. As they conform to the syntactic structures and expressions of Chinese language to render AAVE texts, the distinctive linguistic characteristics of AAVE inevitably disappear in the target texts. Due to this disappearance, readers of the three Chinese translations may not suspect what they read is actually a translated text, especially if the original author's name is also removed. Moreover, it is unlikely for the readers of the three Chinese translations to grasp the linguistic uniqueness of AAVE. Nevertheless, without distorting the original content of the novel, the communicative goal of each Chinese translation is achieved because readers understand the gist of the story, mainly about an African-American woman's struggle for independence.

The readability of the three Chinese translations remains high, but such rendering inevitably results in fluent translation. Lawrence Venuti has warned translator that fluent translation can produce detrimental effect on minimizing the foreignness in the source text. He also argues that foreignization is a non-fluent method designed not only to make translators visible but also to inform readers that the translations originate from a foreign culture. It can take a number of different forms, including regional and social dialects, archaism, jargon and technical terminologies, stylistic innovations and neologisms, literary figures like metaphors. If Venuti's view is taken, the most ideal way for translating literary dialect may well be producing a variation by means of the current local dialect of the receiving language. Are translations rendered by means of foreignization strategy capable of preserving and representing the linguistic features of AAVE in an authentic way?

In her experimental translation of The Adventures of Huckleberry Finn, Ju-yu Lin chooses Taiwanese and Taiwanese Mandarin to translate Jim's black dialect. However, she admits that using written Taiwanese to translate AAVE may cause misunderstanding and problems, even though it is a local dialect with plenty of users in Taiwan. Her experimental translation did not win high praise from either her father or her classmates due to their unfamiliarity with Taiwanese, a spoken language instead of a written one. Therefore, translations into Taiwanese may not reach a high level of readability. She suggests one way to remedy this problem: to supplement the text with explanatory notes, for readability is still a primary goal (2001, p. 38). In light of her attempt, she states that "the feasibility of rendering AAVE in Taiwanese still requires further evaluation" (2001, p. 39). Also, a quantitative study based on questionnaires regarding readers' preference on translations into Taiwanese or the three Chinese translations would be worth carrying out in the future.

Whether the standard or local languages should be employed to translate literary black dialect still remains an unresolved issue. A more effective translation method may be discovered by prospective translators or writers involved in experimental translation and creative writing. Before such a method can be found, it is suggested that translators who choose not to reproduce the distinct syntactic features of AAVE can make good use of a preface or postscript to the translation. Shi's statement about the difficulty of rendering AAVE in her translation's preface has set a good example for prospective translators, although she promises to render the text in plain Chinese owing to the protagonist's language style, Shi does not really introduce the linguistic characteristics of AAVE to her readers. Against this backdrop, one feasible solution may be annotations for each sentence or at the end of each section of the text. However, such renderings may appear redundant in translation. It is instead advised that translator should initiate target readers into the author's intentional use AAVE as well as its linguistic characteristics in a preface or postscript so that readers have a chance to get acquainted with the linguistic uniqueness of AAVE and to appreciate the writer's deliberate use of a particular language style.

¹ There are two different kinds of parallel corpora in existence: uni-directional corpora and bi-directional corpora. As implied by the name, the uni-directional corpus incorporates source texts in language A and translations in language B; the bi-directional corpus consists of four parts, i.e. source texts into language A and translations in language B, as well as source texts in language B and translations into language A.

² Due to the lack of spaces between each Chinese character, software like WordSmith cannot correctly recognize a word. Before processing, three Chinese translations are individually segmented by ICTCLAS (Institute of Computing Technology, Chinese Lexical Analysis System) version 1.0, chiefly designed by Kevin Zhang. See more information at http://www.nlp.org.cn/project/project.php?proj_id=6

Texts