English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Below we develop a method to determine whether the use of grammatical particles in Chinese Buddhist scriptures is characteristic for the period of their translation. The corpus consists of three different Chinese translations of an early Indian Mahāyāna text from two different periods. We use the results of Principal Component Analysis (PCA) to discern if the samples of texts from different periods cluster together. We found that PCA used on common grammatical particles that exhibit great variance between texts, but occur evenly distributed within the texts, does in this case indeed yield distinctive patterns that distinguish the translations from different time periods. This is relevant for historical Chinese linguistics and Buddhist studies. It allows us to identify grammatical particles the use of which differs between translations from different periods. This in turn is an important basis for further research into Buddhist Hybrid Chinese translation idioms and the better attribution and dating of Chinese Buddhist texts. 1. The Texts The Gaṇḍavyūha is an important Mahāyāna text which relates the journey of the merchant’s son Sudhana, who visits 52 teachers in his quest for enlightenment. As usual for the genre we do not have a date or single author for the Gaṇḍavyūha, which started to evolve in Sanskrit probably in Central Asia around the 3rd century CE and was translated * Temple University, Philadelphia, USA ** Dharma Drum Institute of Liberal Arts, Taiwan 1 Chinese in most versions Rufajie入法界 (*Dharmadhātu-praveśana). On the name Gaṇḍavyūha and related terms, see Osto (2008). On the history of the text and its study, in the context of the Buddhāvataṃsaka sūtra see the contributions in Hamar (2007). 2 For an English translation of the G-Tang-1 version see Cleary (1993). Verses from the Gaṇḍavyūha have also been translated in careful comparison with the Sanskrit and Tibetan versions in Gomez (1967). Stylometric Analysis of Chinese Buddhist texts Do different Chinese translations of the Gaṇḍavyūha reflect stylistic features that are typical for their age? Journal of the Japanese Association for Digital Humanities, vol. 2, 2 into Chinese and Tibetan. There are three complete translations of the Gaṇḍavyūha into Chinese: As part of the Buddhāvataṃsaka (T. 278) translated by Buddhabhadra佛陀拔陀羅 et al. in Chang’an between 418 and 420 CE in the final years of the Eastern Jin dynasty (265-420). (Siglum below: G-Jin) As part of the Buddhāvataṃsaka (T. 279) translated by Śikṣānanda實叉難陀 et al. in Chang’an between 695 and 699 CE at the time of the Tang dynasty (618-907). (G-Tang-1) A later, enlarged, Sanskrit version of the Gaṇḍavyūha section was transmitted to China in the 8th century and translated as an independent text (T. 293) by Prajña 般 若 et al. again in Chang’an between 796 and 798 CE. (G-Tang-2) The three full translations represent different stages in the development of the Sanskrit text and differ in length, but agree in general as to their content and structure. There is also an earlier, shorter translation of the Gaṇḍavyūha (G-Qin, T. 294) by the monk Shengjian 聖堅 (fl. 385-422). According to the traditional sutra catalogs Shengjian translated G-Qin between 385-388 in the North-West, during the Western Qin. We will bring G-Qin, which is close in time to G-Jin, into play only later, once we have trained our algorithm on the three full translations. If our method would show that G-Qin and G-Jin are related, it would be an indication that it does distinguish features due to the time of translation. 2. Research Question and Approach Grammatical information in Chinese is expressed through syntax and with the help of a limited number of grammatical particles (xuzi 虛字 ), the use of which changed considerably over time. Our question is: Is there stylometric evidence that the use of grammatical particles in the two Tang versions (G-Tang-1, G-Tang-2) differs from that of the G-Jin version and can the algorithm help to identify which particles make a 3 The main witnesses for the Sanskrit text we have today are based on Nepalese Manuscripts and might thus show the text at a later stage than the Chinese or even the Tibetan versions. The Tibetan (Derge No.44) was translated in the early 9 century. Two short (Buddhist-) Sanskrit fragments that were found in Central Asia, were published in Hori (2002). 4 Strictly speaking, this translation was done under empress Wu Zetian’s short-lived Zhou dynasty, but we will forgo this distinction here. 5 See T.55.2146.119c14 and T.55.2149.254c11. 6 See Gómez (1967: xxiv) for some differences between the abridged early version and the later translation. Another Gaṇḍavyūha text in the canon (T.295) is merely a fragment and cannot be used here. 7 See e.g. the three volumes by Dobson (1959, 1962, 1964) that demonstrate the transition from Early Archaic, to Late Archaic and on to Late Han Chinese by analyzing the changing use of the characters used as particles. For the issue of two-character particles see below. Stylometric Analysis of Chinese Buddhist texts Do different Chinese translations of the Gaṇḍavyūha reflect stylistic features that are typical for their age? Journal of the Japanese Association for Digital Humanities, vol. 2, 3 difference? Can it be shown that the two Tang dynasty translations are stylistically more closely related to each other than to G-Jin? And if yes, which particles are responsible for this difference? Apart from being influenced by the period in which they were produced, the wording of translations obviously depends on the individual style of the translator or the translation team. Traditionally, the language of G-Jin is considered elegant and relatively accessible, and because it represents an early form of the text, it is shorter than the Tang versions. G-Tang-1 is supposedly less easy to read as it retains in places the syntax of its Sanskrit original (Fang 1981: 10 f). The style of the longest version of the Gaṇḍavyūha, G-Tang-2, is considered more fluent (Wen 2000: 19). Content-wise, G-Jin and G-Tang1 are very close, while G-Tang-2 was translated from a considerably expanded Sanskrit text (Gómez 1967: xxvi-xxviii). Traditionally, researchers have studied different translation idioms through close, comparative reading of the texts. For this they either have to limit themselves to a small number of texts or compare only a small number of features. Stylometric analysis can attempt to address longer texts, without the need to sample. Crucially, it is now possible to test a large amount of features, orders of magnitude beyond the reach of the unaided human reader. Two factors must be decided beforehand. First, one has to settle on what type of style markers or features to use; second, which method of comparison to apply. Regarding the latter, current approaches typically combine statistical analysis with machine learning techniques. Regarding the former, generally we try to chose features that appear frequently (“most common words” (Burrows 1992), or “very frequent words” (Hoover 2002)). This works well in many cases, but unfortunately the structure of literary Chinese, whether in its Buddhist or more “classical” variants, is ill suited for a choice of “very frequent words.” In East Asian writing words were traditionally not separated by spaces. Chinese word-segmentation is a difficult subject even for modern Chinese with its better understood dictionary and grammar. The difficulty to disambiguate phrase level compounds and words has been called “one of the most vexing problems in modern Chinese grammar” (Norman 1988: 156). 8 This form of research has produced excellent results for a limited number on texts. Nattier (2008), for instance, gives an overview of the work on the translatorship (de-)attribution for c.50 texts that were translated into Chinese before 280 CE. The arguments have to rely heavily on the catalogue tradition as well as on the comparative use of a relative small number of phrases. 9 The possibilities of stylometric analysis for new discoveries in the field of Buddhist textual scholarship have attracted attention already more than ten years ago when Wan Jinchuan (2002: 73 ff) expressed his hope that stylometrics can decide the doubtful attribution of some Chinese texts to Nāgārjuna. 10 For Chinese and Japanese this is still the case. Vietnamese started to put spaces after each syllable after it adopted the Latin script. Only modern Korean writing, and only since about the seventies, has adopted word segmentation by spaces. 11 This impacts other operations such as PoS parsing or even relatively basic Named Entity Recognition (Bingenheimer 2015). Stylometric Analysis of Chinese Buddhist texts Do different Chinese translations of the Gaṇḍavyūha reflect stylistic features that are typical for their age? Journal of the Japanese Association for Digital Humanities, vol. 2, 4 It has therefore long been held that non-contextual words or phrases are better features for stylometric analysis than words or phrases that human readers tend to focus on. Style is encoded in the use of features such as grammatical particles, basic verbs or even punctuation rather than in the literary use of rare, but “stylish”, nouns and adjectives. Our approach here focuses on grammatical particles (xuzi 虛字), which, next to syntax, are the main carriers of grammatical information in classical Chinese. Apart from deciding on which features to process, stylometrics needs a comparison method that allows to decide on and measure the similarity of features in a multi-dimensional vector space. This type of clustering or classifying problem is currently most often approached with machine learning algorithms, which are either supervised or unsupervised (Stammatos 2009). In supervised methods, such as Bayesian classification (Bozkurt et al. 2007), Support Vector Machines (Diederich et al. 2003), Neural Networks (Tearleet et al. 2007) and others, the algorithm is trained on a set of

journal of the japanese association for digital humanities

Stylometric Analysis of Chinese Buddhist texts - Do different Chinese translations of the Gaṇḍavyūha reflect stylistic features that are typical for their age?