【作者】 夏云；

【导师】 李德凤；

【作者基本信息】 山东大学 ， 英语语言文学， 2010， 博士

【摘要】 本论文是基于语料库方法对英汉翻译小说常规化特征的历时研究。常规化在语料库翻译研究领域通常作为翻译的共性特征提出并得到验证,但翻译作为一种特殊的社会交往行为,社会-文化以及历史的因素不可避免地会在不同语种或历史时期的翻译过程中留下烙印。鉴于此,对常规化特征的历时研究将有助于进一步挖掘其内涵所在,加深对翻译行为本质的认识。本文的主要研究目的是：通过建立英汉历时语料库,描写翻译常规化特征的历时变化,并试图将研究结果放在社会-文化语境中进行讨论,彰显翻译行为的社会文化意义。研究基于以下假设：1)常规化体现为译者倾向于采取目的语常规和典型的表达方式,从而间接导致翻译语言特征的保守性和重复性；2)常规化是一个相对的概念,受翻译规范制约,在不同历史时期,受社会文化语境变化的影响,常规化应有不同的体现方式及程度。翻译语言的常规化可以从语言的表层机制获得外在的体现,而借助语料库手段对翻译文本库进行分析,有助于识别特定时期的常规化或反常规化趋势。历时类比语料库的建设是本研究的起点和基础(第三章)。研究所建语料库包括对应语料部分(DPCECF)和单语类比语料部分(即非翻译汉语文本)。对应语料由14本英语原文取样和两个时期(1930、1940年代与最近20年)的对应汉译文本构成。本文提出了基于语料库的历时研究模式,并从词汇、搭配和句法等层面对不同时期翻译文本的常规化特征进行了多维度的实证分析(第四至六章)。词汇层面的分析主要采用了词性分布、高频词分析、词汇长度以及五个高频词缀的构词能力等方面作为研究的变量。研究结果表明,两个时期的翻译文本在词性总体分布上趋同,并与非翻译文本的词性分布相似,总体上体现了常规化的趋势。译文与非译文的显著差异主要在于名词和代词的使用,名词在两个时期的译文中都趋向于少用,这很可能是由于汉语缺乏形态标记,而用行为动词和描述性形容词来处理原语中大量存在的派生名词的结果,体现了强制性的常规化。代词的冗余则显示了原语的影响。对高频词的分析表明,早期译本相比近20年译本而言,其保守性较强。就词汇长度而言,前期译本的单音节词更为常见,而后期的多音节词有增多趋势。对常用汉语词缀(或准词缀)的构词能力的分析发现,后期译本呈现了较高程度的常规化,而前期译本受原语的影响较大。总而言之,词汇层面的翻译体现了常规化与非常规化特征共存的趋势。对搭配特征的分析表明,前期译本中特定节点词的搭配词较为丰富,“a(的／底)n”搭配模式相比非译本重复率较低,变化性较大,体现了前期翻译的保守性较低。基于对应语料的分析进一步显示,前期译本含有更多的非常规搭配,其原因主要在于对直译手段的偏好,而后期翻译中更常见的处理方法是通过添加修饰语或中心词将原文中的搭配显化,或进行结构上的调整,使其顺应汉语的典型用法。总体而言,前期译本体现了原语较强的影响,常规化趋势弱于后期译本。句法层面的常规化分析主要采用了三个维度：翻译汉语的形合趋势,被动语态的使用,以及句子／句段长度。总体上,翻译汉语的句法特征相比词汇和搭配特征而言,受原语影响更明显,这种影响对于早期译本更为突出,显示了较强的非常规趋势。具体来讲,对形合特征的研究显示,两个时期的翻译文本与非译本相比,都表现出对连接词和介词的高频使用,形合趋势明显。历时比较分析表明,前期译本比后期译本的形合趋势更明显,主要体现为前期译本连词与介词使用更频繁,词形与非译本的重合度较低。此外,前期译本的被动语态频率超出同期的非译本语料,而后期译本被动语态的使用较少于同期的类比语料,表明后期译本更加符合汉语的常规。就被动语态的长度而言,两个时期呈现类似的非常规趋势：受原语影响,长被动的使用少于短被动的比例。而对长被动的施事主语进行检索发现,前期译本受原语较大影响,长度超过三个词的施事主语偏多。两个时期在被动语态的语义韵方面呈现类似的常规化趋势：表示否定语义韵的被动结构占多数比例。句长／句段长的分析表明,两个时期的翻译文本具有共同的反常规化特征,即句子／句段偏长。从历时的角度看,这种反常规化趋势似乎在前期翻译中更为明显,句长／句段长显著高于同期非译本,且趋近于原文本的句长／句段长。进一步分析发现,前置修饰语的频繁使用是句长增加的主要因素,前期译本受原文影响,比后期译本较多使用偏长的前置定语。再者,前期译本与同期非译本相比,定语结构的中心词频繁使用动词和形容词,在一定程度上显示是采用直译手段处理原文中的行为动词和描述性名词所致。针对实证分析的结果,研究认为,不同时期译本的共同特征是译本的杂合性,导致常规化与反常规化特征同时存在于不同时期的译本中。相比而言,早期译本在多数维度上表现出较强的反常规化趋势,造成这种差异的社会文化因素主要包括原语-译语语言文化之间的权力关系,以及由此而引起的语言态度和对翻译策略的观点的变化。20世纪现代汉语白话文的日趋成熟,翻译理论的发展,以及日益增长的语言自信使得翻译过程中对欧化成分的吸收变得日趋理性。“结论”总结了研究的主要结果及其理论与方法论的意义。研究发现,不同时期的翻译特征既有相似又有不同之处,对大多数维度的分析表明,后期译本的常规化趋势强于1930-40年代的译本。其次,不同时期译本都显示,常规化特征与受原语影响而导致的反常规特征并存,体现了翻译的杂合性。再者,句法层面相对于词汇与搭配特征而言,受原语影响较大,非常规趋势较明显。此外,对翻译常规化的研究从侧面表明了原创汉语的“欧化”趋势,这主要是受翻译这一间接语言接触手段影响的结果。总体而言,研究结果与最初的假设大致相符：常规化是一个相对的概念,受社会-文化和历史的因素制约,常规化趋势的变迁体现了翻译行为的社会性本质。本研究所得结论在理论和方法上都对翻译研究有所贡献。理论上,本研究通过对常规化定义、常规化与反常规化、以及常规化与规范／共性的关系等方面的理论探讨,以及多层面的实证研究,有助于揭示翻译常规化的本质,彰显翻译行为的社会性。方法上,本研究提出了常规化研究的历时模式,该模式融合了共时与历时的分析,以及单语类比与双语对应两种语料,有助于扩展翻译语言的研究方法。此外,研究能够帮助我们进一步了解原语、译语与目的语原创语言之间的互动关系,对于研究汉语的发展变迁也有重要启示。但正如其他基于语料库的研究,本研究也存在一些不足之处,包括语料库规模较小,语料标注需要进一步加深等。后续的深入研究中,应注意方法的完善,语料库结构的合理化,标注程度的加深等,以便提高研究结果的可靠性。更多还原

【Abstract】 The dissertation is a corpus-based diachronic research on the phenomenon of normalization in English-Chinese fictional translation. In the field of corpus-based translation studies, normalization has been proposed as a universal feature of translation, however, the notion of "universals" is debatable since translation as a social communicative act will be unavoidably affected by contextual factors in different cultures and historical periods. Therefore, a diachronic study of normalization could shed light on the nature of the notion as well as the nature of translation behaviors. The primary objective of the present research is to systematically explore the variation of normalization across times, and to highlight the social significance of translation activities by contextualizing the research results. Such work is based on the following assumptions:1) Normalization generally refers to the tendency of translators to conform to the conventionally established and standard practices typical of the target language, which would in turn result in a relatively higher conservativeness and repetitiveness, and a lower diversity of language use in translations;2) Normalization is a relative notion, a norm-induced feature that is subject to changes in socio-cultural and historical contexts. Normalization (or de-normalization) is manifest in repeatedly observed patterns in the linguistic make-up of translations as products, and could be discerned using corpus-analysis tools.The construction of a diachronic corpus is where the whole empirical research begins (Chapter Three). The corpus consists of a parallel part (DPCECF) and monolingual comparable part (the corresponding non-translated Chinese texts). DPCECF contains extracts of 14 English source texts with two sets of translations from two time spans: 1930s-40s and the last 20 years. A corpus-based research model is developed for the diachronic study of normalization in translation and a multi-level investigation is conducted on different linguistic levels (Chapter Four to Six).Investigation of normalization on the lexical level uses measures like POS distribution, high-frequency words, word-length, and compositionality of five quasi-affixes. The findings reveal a general tendency toward normalization as regards the general POS distribution of the two translated subcorpora. The translated and non-translated texts differ most evidently in the use of nouns and pronouns. While the lower use of nouns in translations most probably reflects the "conventionalization" tendency in the translation process, the redundant use of pronouns should be due to a major influence from the source English language. Analysis of high-frequency words reveals a stronger conservativeness in TCT1 (translations of 1930s and 1940s) than TCT2 (translations of the last 20 years). TCT1 shows a heavier use of mono-syllabic words, whereas a stronger polysyllabic tendency could be found in TCT2. Study of the compositionality of five quasi-affixes shows a higher degree of translation normalization on the morphological level in TCT2, whereas TCT1 seem to display a stronger influence from the source text language. In sum, analysis on the lexical level finds a mixture of normalization and de-normalization tendencies in the translated subcorpora of different periods.Collocation analyses have found a slightly higher number of collocates in relation to certain node words as well as a lower repetition rate of the "a (的/底) n" lexical patterns in TCT1 in comparison with OCT1 (earlier non-translations). Similarly, parallel concordance data further reveal more frequent de-normalized collocations in TCT1 than in TCT2, displaying a stronger preference for direct translation in TCT1. TCT2 tends to make the original collocations explicit or structurally adapted to typical Chinese usages. Overall, the findings seem to indicate a heavier influence from the source language and a weaker normalization tendency in TCT1 in comparison with TCT2.Analyses on the syntactic level mainly use three dimensions:hypotacticalization of translated Chinese, use of passive constructions, and sentence/sentence segment length. Generally speaking, the syntactic features show more influence from the source language than features on other levels, and earlier translations seem to display a stronger deviance from the normal features of their comparable non-translations. Specifically, as regards the hypotactic features of translated texts, both translated collections seem to show a higher frequency of connectives and prepositions than their comparable counterparts, indicating a stronger hypotactic tendency in translational Chinese. A diachronic analysis shows a stronger hypotactic tendency in TCT1 than in TCT2, as reflected either in the significantly higher frequency of connectives and prepositions in TCT1, or in the lower overlapping rate of word forms between earlier translated and native texts.There is a relative overuse of passive constructions in TCT1 than in the comparable non-translations, whereas TCT2 shows a slight underuse of passives as compared with their counterparts, suggesting a stronger normalization tendency in TCT2. Both TCT1 and TCT2 show a’deviation’from the normal length of Chinese passive constructions, as reflected by a lower frequency percentage of long passives than short passives under the source text influence in the translation process. Within long passives, TCT1 contains a higher proportion of agents over 3 words in length than OCT1 as well as TCT2, exhibiting a stronger influence from English. With respect to semantic prosody, TCT1 and TCT2 tend to be normalized with a major proportion of negative passives in the total cases. The greater length of sentences(S) and sentence segments (SS) suggests a de-normalized feature in translated texts of both periods, and diachronically speaking, this tendency seems stronger in TCT1 with significantly higher S and SS lengths than the corresponding earlier non-translations, and similar mean S and SS lengths to their counterparts in the English source texts. Frequent use of premodifiers is found to be a major factor that contributes to the increase in S length, and further analysis shows a heavier use of long attributives in TCT1 than in TCT2, indicating a stronger influence from the source language on TCT1. Furthermore, the significantly higher number of head verbs and head adjectives in TCT1 than in OCT1 suggests a preference of TCT1 for direct translation of the source text action nouns and descriptive nouns.The empirical research results indicate that hybridity is a common feature in translations of different time spans, and several socio-cultural factors are found crucial in bringing about the different degrees of normalization in translations of different time spans. Generally speaking, change of power relations between source and target languages and cultures leads to the changes in language attitude and views on translation strategies. The gradual development of vernacular Chinese and translation theory, together with the growing linguistic confidence in late 20th century, has made the incorporation of Europeanized language elements more rational.In "Conclusion", the major findings of the entire study are summarized, and the theoretical and methodological implications of these findings are presented. The major findings are as follows:1) Translations of different periods show both commonalities and diversity in the linguistic manifestations of normalization. For most dimensions examined, a stronger normalization tendency could be discerned in translations of recent 20 years than those of 1930s-40s.2) Co-existence of normalization and de-normalization are found in translations of both periods, showing the hybrid nature of translated texts.3) The syntactic features in both translated subcorpora have undergone greater influence from the source language than features on other levels have.4) A Europeanization tendency could be discerned in non-translated Chinese, mainly as result of source language influence in the process of indirect language contact. In sum, the findings confirm our hypotheses that normalization is a relative notion, a norm-induced feature that is subject to changes in socio-cultural and historical context. Change in normalization tendency reflects the social nature of translational behavior.The present research contributes theoretically and methodologically to the fields of translation studies. Theoretically, the study yields insight into the nature of normalization not only with theoretical discussions concerning its definition, its relation with de-normalization, and its norm-induced nature, but, more importantly, with empirical studies on different linguistic levels. This, in turn, will help to unveil the social nature of translation behaviors. Methodologically, the present study proposes a model for the study of normalization in translation, which integrates the synchronic and diachronic perspectives, and combines mono-lingual comparable and bilingual parallel corpus resources. Thus, it expands the research methods for the study of translational language. In addition, the present study helps to offer new insight into the complex, reciprocal relationships between the source language, translational Chinese, and non-translated native Chinese, which has important implications for the study of Chinese language variation. As with all corpus-based research, this research suffers from a number of limitations in corpus size, corpus annotation, etc. Suggestions for further research are proposed concerning more refined methodology, better corpus composition, and in-depth corpus annotation in order to facilitate the reliability the research.更多还原