A Corpus-Based Comparative Multidimensional Analysis of the Two English Translations of Luoyang Jialan Ji ()
1. Introduction
In recent years, many researches have been carried out in the field of multidimensional analysis of English translations based on corpus. He Mengyu conducted a comparative study of the English translation of The Analects of Confucius in five dimensions [1] . In addition, adopting multidimensional analysis to explore the genre features of literary translations [2] and to compare the styles of translators [3] based on corpora are also involved.
Luoyang Jialan Ji “洛阳伽蓝记” is a historical and geographical record of the Buddhist temples in the capital city of the Wei Dynesty (386-534 AD) in ancient China—Luoyang. Written by Yang Xuanzhi (杨衒之, ?-555) in 547 AD, a high-ranking official of the Eastern Wei Dynasty. This classic provides a detailed account of the political events, foreign exchanges, and folk customs over four decades in Luoyang. As a valuable record of the unique political, cultural, and religious life in ancient China, Luoyang Jialan Ji is considered to be one of the most important works in Chinese literature.
Since the 20th century, studies of Luoyang Jialan Ji have primarily focused on its writing style [4] , historical significance [5] , literary influence [6] , ancient garden designs [7] , the author’s experience [8] , social values [9] , and geographical features [10] .
Currently, there are altogether two English translation versions of Luoyang Jialan Ji, i.e., Memories of Lo-yang (ML) [11] translated by W. J. F. Jenner and published in 1981 and A Record of Buddhist Monasteries in Lo-yang (ARBML) [12] translated by Wang Yitong and published in 1984. Therefore, to carry out a study on the English translation of Luoyang Jialan Ji, the above two versions must be carefully analyzed.
Over the past decades, researchers have carried out studies on the English translations of Luoyang Jialan Ji from different aspects, such as the use of lexicons [13] , the translation strategy of folklore events [14] and place names [15] , and the translation of high-frequency words [16] .
However, the existing studies on the English translations of Luoyang Jialan Ji remain incomplete, particularly with a noticeable absence of a comparative analysis of translation styles in the two English translations.
Therefore, by using the MD approach to analyze the two translation versions, this study attempts to compare the most significant similarities and differences of the lexical features from the five dimensions defined by Biber. It is hoped that the results of this study will help future studies on the translations of Luoyang Jialan Ji.
2. Method
This study uses the Multidimensional Analysis Tagger, i.e. MAT, developed by Nini [17] for automatic annotation, feature extraction and data gathering.
The MAT is used to study the two translation versions according to Biber’s multidimensional approach, i.e. MD approach [18] , which is applied to analyzing discourse from several dimensions. The corpus used in this study is two self-built English translation corpora of ARBML and ML, with a corpus size of 46,400 words and 40,075 words respectively.
In this study, the multidimensional analysis includes Dimension 1 to Dimension 5, which are Dimension 1 “Involved versus Informational Production”, Dimension 2 “Narrative versus Non-Narrative Concerns”, Dimension 3 “Explicit versus Situation-Dependent Reference”, Dimension 4 “Overt Expression of Persuasion”, Dimension 5 “Abstract versus Non-Abstract Information” and Dimension 6 “Online Informational Elaboration Marking Stance”. Dimension 6 is not included in this study due to its few features.
The MD approach is efficient in analyzing discourse, especially on the lexical and grammatical levels. More importantly, its classification of discourse dimensions is detailed and matured. Therefore, to carry out studies on discourse analysis, the MD approach is frequently employed.
The overall research procedures are as follows:
1) First, the two translation texts were converted into a TXT format, which were then imported into the MAT application on the computer to generate dimension differences, Z-scores and typical language features, such as AWL, CONC, NN, etc.
2) Next, we started to compare the detailed data generated by the MAT in each dimension classified by Biber [18] . After identifying typical features of the two texts, we then further analyzed the differences and the reasons leading to such differences between them.
3) Finally, we examined the main linguistic features causing the differences between the two texts and summarize the translation characteristics of the two translation versions.
By using quantitative and qualitative analysis through the use of MAT, this study aims to provide a comprehensive understanding of the lexical and grammatical features in ARBML and ML.
3. Results
The results of the MAT analysis are based on the statistics generated by the MD system, which provide details of the comparisons between the two translation versions in altogether six dimensions. Since Dimension 6 is not that related to the book and contains very few important features, it will not be involved in the discussion.
As is shown in Figure 1, there are significant differences between Jenner’s and Wang’s translations in five dimensions. In Dimension 1, both the two versions are more informationally dense than interactional, with Wang’s version being much more informational than Jenner’s. In Dimension 2, both the two versions are narrative, with Jenner’s version being more narrative than Wang’s. In Dimension 3, Wang’s version is relatively more dependent on the text than Jenner’s. In Dimension 4, both the two versions are implicit in their authors’ points of view or assessment, of which Wang’s version contains less subjective viewpoints than Jenner’s. In Dimension 5, there is a positive and negative contrast between the two versions, which indicates that Wang’s version tends to provides information in a more abstract way, whereas Jenner’s in a less abstract way.
![]()
Figure 1. Differences in the five dimensions between the two translations.
Across Dimension 1, 2, 3, 4, a consistent pattern emerges wherein Jenner’s version get higher score than Wang’s. Yet in Dimension 5 there is a contrasting result, which aligns with the narrative and non-narrative differences observed in Dimension 2. In general, texts that are more narrative have a higher density of information, whereas non-narrative texts have a higher degree of informational abstraction [19] .
Statistics demonstrate that Jenner’s ML and Wang’s ARBML exhibit the largest discrepancies in the scores of Dimension 1, Dimension 4 and Dimension 5. Therefore, this study will focus on the typical linguistic features in the five dimensions of the two versions, and summarize their respective translation styles.
4. Discussion
By selecting typical examples in the two translation versions, the following section will have a detailed discussion on the linguistic features of the two translation versions along Biber’s five dimensions.
4.1. Dimension 1
Dimension 1 stands for a dimension marking “high informational density and exact informational content versus affective, interactional, and generalized content” [18] and is called Involved versus Informational Production. The positive score of Dimension 1 indicates interactivity. The higher the score is, the more interactive the text is, for example the casual conversation. The smaller the negative score is, the more informative the text is, like some academic paper. A high score on this dimension means that the text presents many verbs and pronouns (among other features) whereas a low score on this dimension means that the text presents many nouns, long words and adjectives (among other features) [18] .
From the above analysis of dimensions, the two corpora have scores of less than −10 in dimension 1, which indicates that their texts have highly informative features, while Wang’s score is less than −15, indicating that Wang’s ARBML is more informative than Jenner’s ML.
The statistics in Table 1 are the typical linguistic features in Dimension 1 that lead to the differences between the two versions. Among the five dimensions, Dimension 1 contains the most linguistic features such as PASS (Agentless passives), WZPAST (Past participial WHIZ deletion relatives), and HDG (Hedges), all of which share the same scores. There are also features with significant differences such as VPRT (Present tense), POMD (Possibility modals), NN (Total other nouns), PIN (Total prepositional phrases), AWL (Word length), TTR (Type-token ratio). Data JJ (Attributive adjectives) and PIT (Pronoun it) manifest notable distinctions.
![]()
Table 1. Features of Dimension 1 in Jenner’s ML and Wang’s ARBML.
Example 1:
Source Text: 衒之尝与河南尹胡孝世共登之，下临云雨，信哉不虚！
Wang: I once climbed it with Hu Hsiao-shih, the prefect of Ho-nan, and it was abso lutely true that one could look down on clouds and rain. (P23)
Jenner: Hu Xiao-shi, Metropolitan Prefect of He-nan (He-nan yin) and I once as cended the tower. In truth, it seemed as if the clouds and rain were below us! (P151)
Example 2:
Source Text: 唯长乐王子攸像光相具足，端严特妙。是以荣意在长乐。遣苍头王丰入洛，询以为主。
Wang: None but [the statue of] the Prince of Chang-le, [Yuan] zi-you, was dignified or an especially good likeness. As a result, [Er-zhu] Rong was infavor of the Prince of Chang-le. An old and skilled slave, Wang Feng, was sent to enter Luo-yang, requesting that he be the next ruler. (P29)
Jenner: None was successful except that of Tzu-yu, Prince of Ch’ang-lo, which came out as a wonderfully majestic statue, perfect in both likeness and lustre. His mind now set on Tzu-yu, Erhchu Jung sent his slave Wang Feng into Loyang to secure his consent to the throne. (P153)
Example 3:
Source Text: 好事者遂寻文晋朝京师何如今日。
Wang: Out of curiosity, someone then pressed him, asking how he would measure the Jin capital against the present one. (P101)
Jenner: The curious used to follow Chao Yi around and ask him how the Chin capital compared with the modern one. (P183)
The use of PIN (Total prepositional phrases) can realize the simplicity of text and avoid complexity. A preposition and a noun phrase as its complement can realize multiple linguistic functions.
Example 1, 2 and 3 all show that Wang used more PIN than Jenner did. In Example 1, Jenner used “it was absolutely true” to describe what he saw after climbing the tower, while Wang used PIN “in truth” to express the same meaning, expressing his feelings with concise language.
In Example 2 “是以荣意在长乐” was translated into “as a result, Er-zhu Rong was in favor of the Prince of Chang-le” and the prepositional phrase increased the information of Wang’s ARBWL. Jenner translated it as “his mind now set on Tzu-yu,” thus omitted the logical relationship between the sentences.
In Example 3, Jenner used “the curious”, whereas Wang employed “out of curiosity”. The latter comprises an abstract noun combined with the preposition “out of”, making the plot vivid and easier for readers to empathize with it. It also avoids monotonous wording, enhancing the richness of the translation through varied expressions. The using of prepositional phrases makes Wang’s ARBWL more informative.
Example 4:
Source Text: 于时新经大兵，人物歼尽，流迸之徒，惊骇未出。
Wang: This was a period when the nation had just undergone severe military disturbances, resulting in the liquidation of many dignitaries. Those who had fled were too frightened to appear. (P31)
Jenner: After the recent heavy fighting in which all the leading personalities had been exterminated those who had fled were still too frightened to come out of hiding. (P154)
Example 5:
Source Text: 加荣使持节中外诸军事大将军、开府北道大行台、都督十州诸军事大将军、领左右、太原王。
Wang: More honorific titles were given to [Er-zhu] Rong and [Yuan Tian-]mu: for [Er-zhu] Rong… (P31)
Jenner: Erhchu Jung was appointed Senior General Controlling Domestic and For eign Military Affairs… (P154)
Nouns are the main means by which writers refer to specific entities or a set of concepts, and their high frequency reflects a high density of information [1] . In Example 4, Wang emphasized the time using “this was a period”, while Jenner directly described the situation of the war. In Example 5, Jenner used characters as the subject, and Wang used “honorific titles” to explain it first, which increased the information of Wang’s ARBWL. Summative noun “honorific titles” helps the reader quickly understand that the translation introduces the titles of Erhchu Jung without having to read a long paragraph of text, which can be captured at the beginning of the sentence, the same as the using of “period”.
4.2. Dimension 2
Dimension 2 is the opposition between Narrative and Non-Narrative Concerns [18] . It can be considered as the one to distinguish between active, event-oriented discourse and more static, descriptive or expository types of discourse [20] . Lower scores on this dimension suggest that the text is non-narrative in nature, while higher scores indicate a narrative text, such as a novel. A higher score on this dimension implies that the text contains numerous past tenses and third-person pronouns among other distinguishing features.
For Dimension 2, past tense and third person pronouns are important linguistic features, and the high frequency of these two types of words represents high narrative property of a text.
As is shown in Table 2, TPP3 (Third person pronouns) is an important linguistic feature in Dimension 2.
![]()
Table 2. Features of Dimension 2 in Jenner’s ML and Wang’s ARBML.
Example 6:
Source Text: 中有丈八金像一躯，中长金像十躯，绣珠像三躯，金织成像五躯，玉像二躯。作工奇巧，冠于当世。
Wang: In the hall was a golden statue of the Buddha eighteen Chinese feet high, along with ten medium-sized images-three of sewn pearls, five of woven golden threads, and two of jade. The superb artistry was matchless, unparalleled in its day. (P19)
Jenner: In it were an 18-foot-high gold statue, ten man-sized gold statues, three statues studded with pearls, five statues woven from goldthread, and two jade statues. They were all of brilliant and unmatched workmanship. (P149)
In Example 6, Jenner used “it” and “they” instead of specifying the subject “hall” and “superb artistry”, and the massive uses of third person pronouns make Jenner’s ML more narrative. English tends to use more pronouns to convey logical information by reference; Chinese tends to repeat nouns or names, using retelling to convey logical information [21] . However, the anaphora of English third person pronouns has a wide range and flexible application, which can not only enhance the cohesion of discourse, but also realize the improvement of narrative.
4.3. Dimension 3
Dimension 3 is defined as “Explicit versus Situation-Dependent Reference” [18] . A high score in this dimension indicates that the text is less or not dependent on the context, whereas a low score indicates the text is dependent on the context.
As is shown in Table 3, in both Jenner’s ML and Wang’s ARBML, RB (adverb) and NOMZ (nominalization) are negative features, of which, Wang’s ARBML contains more nominalizations than Jenner’s ML. This feature is reflected in Example 7. However, Jenner’s total score of the linguistic features in Dimensions 3 is higher than Wang’s, which indicates that Jenner is less dependent on the context than Wang.
![]()
Table 3. Features of Dimension 3 in Jenner’s ML and Wang’s ARBML.
Example 7:
Source Text: 部落八千馀，家有马数万匹，富等天府。
Jenner: He had over 8000 families of tribesmen, several tens of thousands of horses, and wealth to match a heavenly treasury. (P152)
Wang: He was in possession of scores of thousands of horses, and his wealth equaled that of the vassal of a rich kingdom. (P25)
Nominalization is one of the most important features in Dimension 3. Generally speaking, the more nominalizations a text uses, the less dependent on the context the text is. However, statistics demonstrate a reserve result in Jenner’s and Wang’s text. In Example 7, Jenner used a verb “had” to describe the man’s poverty, while Wang employed a noun phrase “in possession of” to achieve the same purpose. As is shown in Example 7, Jenner’s translation presents less nominalization than Wang’s.
4.4. Dimension 4
Dimension 4 is defined as “Overt Expression of Persuasion” [18] . A high score in this dimension indicates that the text explicitly marks the author’s points of view and assessments of possibility, which presents more modal verbs than other features, whereas a low score indicates the author of the text tends to express his or her perspective of a certain thing, person or event in a more implicit way with less modal verbs.
As is shown in Table 4, in overall, Jenner’s ML contains more MD (modal verbs), especially possibility modals than Wang’s ARBML. The core of modality is that it reflects the attitude and judgment of the author or speaker [22] . Thus, this feature indicates that Jenner shows more overt persuasive effort in his translation.
![]()
Table 4. Features of Dimension 4 in Jenner’s ML and Wang’s ARBML.
Table 5 depicts the differences of the frequency of the possibility modals in Dimension 4 between Jenner’s ML and Wang’s ARBML. Biber [18] listed four possibility modals in his book: can, may, might, and could. In overall, Jenner’s ML contains more possibility modal verbs than Wang’s ARBML.
![]()
Table 5. Frequency of possibility modals in Jenner’s ML and Wang’s ARBML.
In Jenner’s ML, there is a high frequency in the use of “could”. This model verb is mostly used to describe ability in Jenner’s translation, followed by possibility.
Although “could” is also the most frequently used model verb in Wang’s translation, its raw frequency is much lower than Jenner’s. This may due to Wang’s balanced use of other model verbs, such as “can” and “may”.
The two following examples show the features of the use of “could” in Jenner’s and Wang’s translations.
Example 8:
Source Text: 衒之尝与河南尹胡孝世共登之，下临云雨，信哉不虚！
Jenner: I once climbed it with Hu Hsiao-shih, the prefect of Ho-nan, and it was absolutely true that one could look down on clouds and rain. (P151)
Wang: Hu Xiao-shi, Metropolitan Prefect of He-nan (He-nan yin) and I once ascended the tower. In truth, it seemed as if the clouds and rain were below us! (P23)
In Example 8, Jenner used the model verb “could” to describe that one has the ability of seeing the clouds and rain below the tower. Yet, Wang employed a predicative clause to express the same meaning. This discrepancy indicates that Jenner tends to explicitly mark his point of view, while Wang tends to describe things from a more objective perspective by hiding the actor of the action within the text.
Example 9:
Source Text: 虎贲张车渠，掷刀出楼一丈。
Jenner: He could throw a spear as high as a 100-foot tree. The Tiger Guardsman Chang Ch’e-chu could throw a sword 1 chang higher than a tall building. (P253)
Wang: He could throw a halberd as high as the top of a tree one hundred Chinese feet above the ground. (P257)
In Example 9, both Jenner and Wang use “could” to describe the man’s ability of throwing a spear or halberd. It is conspicuous in the two translation versions that both Jenner and Wang prefer using “could” when the intension is to underscore a special capacity that belongs to a certain individual. “Could” is more often used to emphasize something that one person is able to do, rather than things that can be done by individuals.
4.5. Dimension 5
Dimension 5 is the opposition between Abstract and Non-Abstract Information. High scores on this variable indicate that the text provides information in a technical, abstract and formal way, as for example in scientific discourse. A high score on this dimension means that the text presents many passive clauses and conjuncts (among other features) [18] .
Table 6 illustrates that ARBML got a positive score of 1.25 on CONJ (Conjunction) while ML got a negative one.
![]()
Table 6. Features of Dimension 5 in Jenner’s ML and Wang’s ARBML.
Example 10:
Source Text: 荣为盟主，穆亦拜荣。
Wang: Yuan Tian-mu was senior in age, so by convention he was treated as the elder brother with the accompanying respect. Er-zhu Rong, however, was recognized as the sworn leader, so that Yuan Tian-mu bowed to the other for courtesy’s sake. (P27)
Jenner: Erhchu Jung and Yüan T’ien-mu then swore brotherhood. As T’ienmu was the older Jung honored him as the elder brother; and T’ien-mu bowed to Jung as the leader of the alliance. (P152)
In Example 10, Wang paid more attention to the logic of sentences than Jenner. “穆亦拜荣” was translated into “Er-zhu Rong, however, was recognized as the sworn leader, so that Yuan Tian-mu bowed to the other for courtesy’s sake,” which uses the CONJ (Conjunction) “however” to represent the transition relationship of sentences, thus making Wang’s ARBML more abstract and get a higher score in Dimension 5.
Wang, as a Chinese translator, may lay more emphasis on the differences between Chinese and English in his translation. English focuses on hypotaxis and emphasizes form and logic. Chinese attaches great importance to parataxis, which does not use too many linguistic forms to connect the sentences, but reflects its grammatical meaning and logical relations through the meaning of words or clauses.
Example 11:
Source Text: 里内有京兆人杜子休宅，地形显敞，门临御道。
Wang: His residence faced the Imperial Drive and was spacious and enjoyed an inspiring location. (P99)
Jenner: It was set in spacious ground and its gates opened on the imperial highway. (P182)
In Example 11, Wang used “his residence faced” while Jenner used “it was set”. The frequent use of PASS (Agentless passives) making Jenner got a higher score. It also takes into account that as a translator in the source language’s country, Jenner is better at using the passive voice, which also reflects the difference between Chinese and English.
Discourses with very frequent passive constructions are typically abstract and technical in content, and formal in style. Apparently, conjuncts and adverbial subordinators frequently co-occur with passive forms to mark the complex logical relations among clauses that characterize this type of discourses [20] . Although the statistic of PASS (Agentless passives) should be proportional to the score of abstraction in Dimension 5. However, the positive influence of other linguistic features is greater than that of PASS (Agentless passives), so Wang’s ARBML is still more abstract.
5. Conclusions
After analyzing Jenner’s ML and Wang’s ARBML with the help of the MD approach, this study tries to summarize the styles of the two translation versions of Luoyang Jialan Ji and their respective characteristics.
As can be seen from the above discussion, there are significant differences between the two versions mainly in Dimension 1, 4, and 5. Wang and Jenner present different translations of Luoyang Jialan Ji to readers from different perspectives. As a local translator, Wang has a deep understanding of traditional Chinese culture. On the basis of accurately conveying the meaning of the text, he pays more attention to the function and role of the text, so he is more prominent in the aspects of information and abstract. Jenner, on the other hand, can add narrative, situational, and persuasive elements to better understand the target language, making it easier for native English speakers to read.
In the process of research, we have the following inspiration. The MD approach of conducting multidimensional analysis can greatly improve the efficiency of studying texts like Chinese classic works, helping readers acquiring a clear understanding. Perhaps it can also be combined with grammar learning and translation teaching to break the restrictions of textbooks, teachers and machine translation. The dimensions and linguistic features obtained after the text is imported into the corpus can show the characteristics of the text in a direct way, providing students with a convenient way of learning. Teachers can guide students in analyzing statistics to discern the differences between Chinese texts and English texts in the respects of vocabulary, syntax and rhetoric. To explore translation strategies and methods, we can shift the focus of translation teaching from theoretical knowledge to independent learning. The MD approach can play an important role in both translation studies and practice, providing perspectives that cannot be found by reading alone.
For similar studies in the future, we believe that researchers can choose other corpus analysis applications such as Antconc or Lancsbox for quantification and analysis, so as to obtain translation differences from multiple perspectives. Quantitative analysis tools such as SPSS can be combined to explore the difference and regression between the translations.
In terms of selecting research objects, scholars can also expand their focus beyond literary works and classics. Other materials such as historical documents and religious classics can be explored by using similar methods to further promote intercultural communication.
However, this study also has some limitations. It is challenging to summarize the translation styles of two translators through a single text, and the differences in linguistic features may be intricately linked to the different cultural backgrounds and translation purposes of the translators. To address these challenges, we may need more examples and information for further studies.
Acknowledgements
This study is supported by the Undergraduate Training Program on Innovation and Entrepreneurship grant 202310251103, which is a national level program. Qin Gu and Ling Chen would like to express our sincere gratitude to the program that has offered us this opportunity to carry out the study.