| Translating, the process of providing a written rendering of a text which is both true to the content of the original donor text and clear and idiomatic in the receptor language, is a complex and fascinating task. In fact, Richards (1953) has claimed that it is probably the most complex type of event in the history of the cosmos.One pursuit that is closely related to translating and probably can date as long ago is the translation teaching. In recent years, significant advances have been witnessed in this area including the teaching methodology, the course development and the textbook compilation. As an important component of translation teaching, translation testing, though, still lags far behind its counterparts with regard to its theory and practice, therefore undermining the assessment of the effectiveness of translation teaching as well as students'translation achievements. As a matter of fact, this situation reflects a general plight in translation quality evaluation, the umbrella of which translation testing falls in. According to Horguelin (1985), translation quality evaluation has long been a very subjective exercise, but in the contemporary period the appreciation of quality in translation has ceased to be a debate based on general, vague criteria and has focused on a search for more objective systems of evaluation, ranging from simple value scales to sophisticated global models. However, as Melis and Albir (2001: 274) argue, there is still no operative model of analysis for translation evaluation that has been sufficiently validated by means of empirical-experimental research.One of the problems of translation testing lies in their construct validity, that is, the extent to which they are based on an appropriate model of translation competence. As Campbell (1998: 8) points out,"I would argue that most translation quality assessment systems are not underpinned by theory at all, at least in any principled or even conscious way on the part of their designers". The observations have been shared by some Chinese translation scholars, who started to concern themselves with the issue back in 1980s and 1990s. Mu Lei, one of the most prominent translation scholars in China, made her point clear recently,"There has been a lack of systematic research into translation testing. The use of objective items in translation testing is at its infant stage; therefore the subjective tests constitute the main stream in translation assessment. Even though there are a variety of rating schemes for the format, none of them has been subjected to rigorous theoretical probing; the translation testing approach, textual procedures in particular, is saddled with unreliability and haphazardness"(2006a: 468). She also made the following comments on the translation testing research,"Given the currently available literature, there has been a dearth of investigations dedicated to translation testing, let alone more systematic and theoretically-grounded research. Interestingly, quite a large number of people are keen on addressing general translation quality evaluation, which is closely related to translation testing; it is a pity that the majority of the research is pure theoretical in nature and do not lend themselves to the application by frontier teachers, which highlights the importance of the research directly into translation testing"(2006b: Foreword).Thus, a model of translation competence does appear to be essential to translation testing schemes. Alderson, in the case of a language test, made the point clear that"every test has a theory behind it: some abstract belief of what language is, what language proficiency consists of, what language learning involves and what language users do with language"(2000:16). Though a translation test is distinct from a language test, the idea is equally applicable here; a set of beliefs or explanations about what makes a good translator must inevitably be the blueprint for any respectable translation testing procedure.As a matter of fact, there is no shortage of translation competence models in the translation academia; various scholars have put forward their own versions of the construct in question, notably Wilss (1982), Bell (2001), Newbert (2000), PACTE research group (2000; 2003) in the West and Yang Xiaorong (2002), Jiang Qiuxia & Quan Xiaohui (2002), Liu Miqing (2003), Wen Jun (2004) in China, among others. What is in common among these models is that the notion of translation competence has been viewed as a multi-component construct involving sets of subskills such as linguistic, cultural, communicative, encyclopedic, transfer, and, in most cases, unprecedented notice has been given to strategic manipulations. It is true that these models deliver an in-depth and comprehensive reflections on the translation competence, but they are also"theoretical and speculative rather than empirical and were based on idealizations rather than on actually occurring data"(Toury, 1981: 41). From the perspective of translation teaching, likewise, these works tend to be expert- or theory-centered, rather than student-centered."Their prescriptive nature automatically prevents them from being able to describe competence per se; what they describe is the anticipated results of the programmes that they espouse"(Campbell, 1998: 11). Recently, though, there have appeared a handful of translation competence models targeted directly at student translators, as proposed by Campbell (1998), Hai Fang (2004) and Zhu Yubin (2008), yet their theoretical underpinnings and usefulness in the translation testing practice still has to be empirically validated.Given the above background, the author of the current dissertation sets to further explore translation testing in the hope of shedding more light on the issue. Specifically, the focus has been sharpened to the student translators --- senior English majors in China --- working into English, a second language. As the students play a dual role --- both a translator and a learner --- and any translator working into a second language is, in reality, on a developmental path with respect to that language, the efforts to understand their Chinese-to-English (C-to-E) translation competence and the relevant testing schemes can be considered from an interlanguage perspective, which calls for an innovative endeavor while tapping into the previously reviewed translation competence models. Moreover, the current research firmly establishes itself in the context of high-stake summative tests represented by the C-to-E translation section (with the task of requiring the test takers to complete the translation into English of a passage of approximately 150 Chinese words within 30 minutes) of Test for English Majors Band 8 (TEM-8), which is a key instrument in China to examine the effectiveness of the implementation of its National Teaching Syllabus for senior English Majors.The nature and the composition of the construct of the C-to-E translation competence for English-major students has been a major focus of attention in this dissertation. Another key focus of the attention has been the study of C-to-E translation assessment criteria and the redevelopment of the rating scale for TEM-8. To be more precise, the research questions the current dissertation attempts to address are formulated as follows:1. What does the translation competence of China's senior English majors mainly consist of against the background of high-stake textual translation task (eg. the C-to-E translation section of TEM-8)? What represent the most apparent and differentiating elements in the construct of the C-to-E translation competence? Is it possible to put together a more empirically-grounded and applicable model of translation competence targeted at those students?2. To what extent can the task of C-to-E textual translation, based on the prospective model of the translation competence, lend to a rapid and effective evaluation of their performance in question? This can be subdivided into two branch research questions, ie.,a What sort of adjustments may be needed for the current rating scale and criteria for the C-to-E translation section of TEM-8?Is it possible to draft up a new marking scheme informed by the current research?b How reliable and valid is the new marking scheme, if formulated? Is it fully operational in the test practice and what impact it might have on the source text selection for the task of C-to-E textual translation?In terms of the theoretical rationale, the current research draws primarily upon the modern language testing theories which center on the concepts of construct and construct validity of a test, and, to be more exact, it is largely enlightened by the contemporary views that conceptualize the construct of competence and the design and marking of performance tests into a symbiotic, well-balanced and interactive relationship. On one hand, the development and application of a valid performance test must solidly build upon a clearly defined and well spelt-out construct of the competence, as is noted by McNamara (1996) that the scale that is used in assessing performance tasks represents, implicitly or explicitly, the theoretical basis upon which the test is founded: that is, it embodies the test (or the scale) developer's notion of what skills or abilities are being measured by the test. On the other, the establishment of the construct cannot go without the specific context because"the construct is not a psychological entity that exists in our mind but an abstract definition made for a particular testing purpose"(Alderson, 2000: 118)."No single definition of a language ability is applicable to all situations; rather, for each testing situation, a definition of the ability, or construct, of interest must be developed that takes into account the test takers, the purpose of the test, and the target language use situation". (Weigle, 2002: 41-42).In terms of the methodology, the current research is empirical in nature, based on observations and experimentations with the analytical approaches drawing upon the areas of language testing and applied translation research respectively.From the perspective of language testing, measurement is at its core and therefore it is empirical in essence; despite the fact that the definition of a language ability is closely associated with the subjective interpretations of its researchers, the inspection of its integrity as well as the consequent reliability and validity of the ability-based test cannot afford to be purely subjective. It has been generally acknowledged in the language testing community that the development and subsequent revision of a rating scale and descriptors for each scale level are of great importance for the validity of any assessment in that raters are expected to make decisions on the basis of common interpretations of the scale contents; moreover, criteria and rating scales, as well as content and format, need to be kept under review during the operational life of any test and are likely to require revision from time to time. Traditionally, there has been a tendency for rating scales to be a priori measuring instruments, that is, their development has been dependent upon the intuitive judgment of an expert. Recent research, however, has suggested a more empirically-oriented approach to the generation of rating scales through the examination of actual scripts and analysis of samples of actual language performance constituted an integral part of the rating scale and criteria development stage (Shohamy, 1990; Upshur & Turner, 1995; Milanovic etc., 1996; Fulcher, 1996; Hawkey & Barker, 2004). Accordingly, the current research sets to conduct an in-depth analysis of the English majors'actual C-to-E translation scripts provided by the related corpora in a bid to better understand about key features of the translation ability across different proficiency levels.The last few years have seen a much clearer identification and understanding of the many different avenues which are open to researchers in the field of translation research. More studies are now resorting to descriptive approaches than to prescriptive approaches; the development of new and sophisticated qualitative and quantitative methods is facilitating our investigative and analytical research work, with corpus linguistics, verbal protocol analysis, statistical tabulation are just a few examples. Apart from the relative new field of corpus linguistics, the instruments such as think-aloud protocols also offers considerable promise in the translation research and the probing into the translation process in particular. In a nutshell, different methods are used in this research to collect data about the process (experimental) and the product (corpus) and the results are triangulated.The dissertation per se begins with a comprehensive review of construct-centered modern language testing theories, the exploration of the notion of translation competence by the Chinese and international translation academia and the status quo of the current practice regarding the translation testing across the world with particular reference to their rating schemes. The author argues that the numerous translation competence models mentioned here either are targeted at translation professionals or fail to be put in direct use in the testing procedure. The two models, though, offer tremendous insights into the current research. Placing the students'translation into the second language in the framework of second language acquisition, Campbell (1998), through an empirical investigation into the translation scripts of 38 candidates entering for the public translation test in Australia, proposes a model for the L2 translation competence which consists of three components, namely, target language textual competence, disposition of the translator and the monitoring competence. However, the author challenges Campbell's work from two lines: an overemphasis on the student translator's target language textual competence and a lack of detailed illustrations as to how his model may guide the translation testing. In fact, the author of the current dissertation believes that except for the textual competence, the subsets of the disposition and monitoring competence are hard, if not impossible, to be assessed especially in a one-go summative test of high stakes. Hai Fang (2004) devotes her research into the translation testing theory and practice with reference to English majors in China, which is more in line with the subject of the current research. By means of partially empirical work, she hypothesizes that the translation competence of English majors in China is composed of an advanced source language reading competence, a translation strategy competence as well as an advanced target language writing competence. It should be noted that her model is of dynamic nature and better grounded in the reality of the English majors here in China. Still, two questions have been raised by the author of the current dissertation: first, as the translation strategy is a purely psychological trait, it might be tricky to subject it to relatively objective assessment; second, the subdivision of the translation competence into the reading, strategy and writing components entails the corresponding evaluation in the above three regards, which might impose a heavy burden on the real testing procedure, thus calling its practicability into question.The retrospection of the current translation testing practice reveals that a great variety of rating schemes--- error deduction approach, impression approach, comprehensive approach, analytical approach--- have been in use; yet hardly any of them is based on a well-grounded notion of translation competence. The literature about language testing also indicates that although analytic schemes may facilitate the agreement amongst examiners as to the precise range of qualities that are to be evaluated in an essay, the actual amount of individual marker variation, ie., degree of unreliability, in many schemes may be reduced very little if there is a lack of explicitness with regard to applicable criteria, or a use of vague criteria. Similarly, Hamp-Lyons (1990) argues that the raters should focus only on the most salient criteria as established through careful test development, well grounded in actual data and in the context where measurements are to be made. As part of the background-setting exercise, the dissertation also reports the result of a questionnaire survey which was completed by approximately 50 raters of the C-to-E translation section of 2009 TEM-8. The purpose was to elicit from the raters their individual approaches and attitudes to the translation assessment. According to the survey, the involved raters generally believed that there is a certain degree of discriminability in the current five-band rating scale for the C-to-E translation section of TEM-8 whereas its interpretability and operationability is yet to be enhanced, hence the ratings are prone to the influence of grammatical errors, local translation quality in a script and the impositions of individual raters. All this illuminates the importance of a new rating scheme underpinned by an empirically hoisted model of students'C-to-E translation competence.The dissertation then embarks upon seeking empirical data that could be employed to induct the model in question, which constitutes one of the core components of the research. The learners'translation corpus with the aid of concordance tools, as a combination of qualitative and quantitative techniques, can help reach a consistent and reliable analysis with translation errors and types identified and tabulated. Tapping into approximately 2000 C-to-E translation scripts from 2004-2006 TEM-8 which naturally makes up a learners'translation corpora, the author conducts an overall comprehensive as well as a sampling discriminative analysis, obtaining a cross-section picture of the translation performance by senior English majors in China. The analysis indicates that in addition to grammatical errors, some key patterns of translation errors also emerge in the students'work, such as the inexact rendering of a source language word, redundant expressions, Chinglish sentence structures, a lack of cohesive devices in the target language, amongst others. It is hypothesized that these patterns can be attributed to the translation errors at two levels– lexical and syntactical transference, which might be the crucial differential features across students of different C-E translation competence. A student of higher C-E translation competence may not necessarily come up with a better rendering in a particular point of observation than the one of lower C-E translation competence; however, the former outperforms the latter significantly as a broader area of observation points are concerned.During the last few years, many arguments have been put forward about the advantages and disadvantages of using think-aloud protocols to investigate the subjects'mental processes. Still, it can be assumed, as the author claims, that this procedure in combination with retrospective probing represents a useful instrument to formulate hypotheses on the cognitive activities during the translation processes. The dissertation, therefore, sets out to report on the results of a TAP experiment on twelve fourth-year English majors of different C-E translation competences, manifested by their translation teachers'judgment and the results of the translation courses they have taken. The subjects were asked to verbalize the thoughts that occurred to them as much as possible while they performed the translation of two short Chinese passages. The exploitation of this procedure has delivered fresh and exciting insights into the nature of the student's C-to-E translation competence. It has been found that there are considerable variations among them as to the decoding of the source text, treatment of the translation unit, mental readiness of adding cohesive devices, simplifying redundant versions and in-depth sentence restructuring in the target text; the same is true of producing truly equivalent words that match the source text words in reference, style, collocation. The author further argues that these variations roughly correspond to their transference processing at the level of lexis and syntax with high level students demonstrating a much clearer sign of breaking the shackles of word-for-word translation and deep-level restructuring of the target text.Based on the above data collection, the dissertation proceeds to propose a model of students'C-to-E translation competence against the TEM-8 background. The model centers on the student translators'awareness of inter-lingual meaning realization and conceptual structure transference and the extent to which they can break the shackles of the source text and undertake an in-depth decoding and recoding exercise during the translation process. Specifically, it is made up of three components --- C-to-E linguistic subcompetence, C-to-E lexical transference subcompetence and C-to-E syntactical transference subcompetence. The dissertation consequently puts forward a new marking scheme for the C-to-E translation section of TEM-8, which, as opposed to the current one, is analytical and dynamic in nature, consisting of grade-specific overall description, grade-specific operational description and a text-specific collection of observation points. The grade-specific overall description roughly corresponds to the current TEM-8 rating scale and criteria, providing a bird's-eye-view judgment of the translation works'fidelity to the source text, their naturalness and idiomaticity as well as grammatical accuracy. The grade-specific operational description, which generally matches the above C-to-E translation competence model, elaborates on the construct of fidelity, idiomaticity and grammatical accuracy, facilitating a clearer, better spelled-out judgment on the part of the raters. And the text-specific collection of observation points enables the adjustment of marking operations to the individual source text chose for each year's TEM-8 with the possible result of a more flexible, operational and better targeted rating exercise.Finally, the dissertation demonstrates the feasibility of the new marking scheme through a trial marking of five students'C-to-E translation scripts from 2005 and 2006 TEM-8. The initial reliability of the new marking scheme has also been proved as a result of the correlation analysis of the rating scores on 150 sampling scripts by two new-scheme-informed raters; so has the initial validity of the new marking scheme through a triangulated correlation of the new rating scores, TEM-8 rating scores and the assumed expected rating scores (the means of the scores respectively given by well-trained two raters based on the current rating criteria ) on the above 150 scripts. In addition, the dissertation also briefly addresses the possible impact of the new marking scheme on the source text selection of TEM8's C-to-E translation section and the undergraduate C-to-E teaching for English Majors.In the final analysis, the dissertation is of a certain amount of academic significance in that it facilitates a more thorough understanding of the C-to-E translation competence of English majors in China and may illuminate a development path of the competence throughout their undergraduate years. Also, the suggestion of a new marking scheme for the C-to-E translation may enhance the backwash effect of TEM-8 by prompting the improvement of the validity of the relevant section. Over the long run, the framework and methodologies used in this research might be extended to the investigation of the English-to-Chinese translation and other aspects of the applied translation studies.
|
PDF Full Text Request