"In this book, Richard Kern explores how technology matters to language and the ways in which we use it. Kern reveals how material, social and individual resources interact in the design of textual meaning, and how that interaction plays out across contexts of communication, different situations of technological mediation, and different moments in time."
Review of Cross-Linguistic Variation in System and Text
Date: Mon, 2 Feb 2004 12:13:19 +0000 From: Gabriela Saldanha Subject: Cross-Linguistic Variation in System and Text
AUTHOR: Teich, Elke TITLE: Cross-Linguistic Variation in System and Text SUBTITLE: A Methodology for the Investigation of Translations and Comparable Texts SERIES: Text, Translation, Computational Processing PUBLISHER: Mouton de Gruyter YEAR: 2003
Gabriela Saldanha, Centre for Translation and Textual Studies, School of Applied Languages and Intercultural Studies, Dublin City University.
The aim of this book as stated in the Introduction (Chapter 1) is to develop a methodology for the contrastive analysis of translations and multilingually comparable texts based on the lexico-grammatical features of those texts. In Chapter 2 the author describes the most recent work on cross-linguistic variation in translation from the perspectives of language typology, contrastive linguistics and translation studies. Teich offers a critical review of Hawkins' comparative typology of English and German (1986), Doherty's research on English-German contrasts in translation (1991,1993), Baker's universal features of translations (1995,1996) and Biber's cross-linguistic comparison of register variation (1988, 1995).
Based on the insights provided by such works Teich establishes the requirements for a new methodology that focuses specifically on the analysis of cross-linguistic variation in translations and comparable texts. This methodology starts from the assumptions taken from translation studies - specifically from Baker's work - according to which translations have specific properties that distinguish them from original texts. The aim of the methodology is answering the research questions arising from those assumptions, namely, which are those properties and what are their underlying motivations. The analysis is restricted to one language pair (German and English) and the features to be analysed are selected from typological studies and register analysis in the two languages. The hypotheses are formulated on the basis of a model of cross- linguistic variation based on Systemic Functional Linguistics (SFL). The analysis is then carried out using corpus-based, quantitative techniques. The results are interpreted in light of the contrastive account of commonalities and differences in the two linguistic systems based on the model of cross-linguistic variation proposed.
In Chapter 3 the model of cross-linguistic variation is developed. Because this model is based on Systemic Functional Linguistics (SFL), this chapter first discusses the main theoretical assumptions and the core components of SFL. This description is quite detailed and focuses on the six major representational categories of SFL: stratification, metafunction, rank, axis, delicacy and instantiation. The model of cross linguistic variation which is then proposed simply outlines the possible variations across language systems in each of the six dimensions. So, for example, regarding metafunctions it is established that different languages may distribute functional responsibilities differently across metafunctions, and regarding instantiation, that even when there is a basic commonality between two languages in a particular grammatical system, the instantiations of the grammatical system may differ, i.e. the two languages may still have different choice preferences according to situational context (register). Examples of variations across languages in each of the six representational categories are then described for clarification. Finally, the advantages of this model for the study of language typology and translations are highlighted and two general hypotheses regarding the effects of cross-linguistic variation on translations are formulated.
Chapter 4 offers a detailed account of grammatical contrasts and commonalities between English and German. The description focuses on the semantics of the clause and their grammatical realisation. At the experiential level the systems of transitivity and voice are described, at the interpersonal level, the systems of mood and modality are described, and regarding the textual metafunction, the theme/rheme and information distribution (new/given) options are described for each language. The main contrastive features of English and German regarding the nominal group, the prepositional phrase, the verbal group and morphology are also discussed.
Both the description of the systemic functional model of language and of the English and German linguistic systems are very detailed and assume no previous knowledge on the part of the reader. While having the advantage of making the book accessible for a wide audience, which, given the interdisciplinary nature of the book, is highly desirable, the high level of redundancy and technical detail makes these two chapters relatively long and dense, especially taking into account that not all of the information provided will be used in the subsequent analysis and interpretation of results (there are hardly any references in these sections to the systems of mood and modality, for instance).
In Chapter 5 hypotheses regarding the nature of translated texts are formulated and tested. Two hypotheses are first established on the basis of two tendencies observed in translated texts by a number of scholars: that of source language (SL) interference, which Teich calls SL shining through, and that of target text (TT) normalisation. According to the first hypothesis the lexico-grammatical properties of the SL, as evident in a particular register, may shine through in the TT. According to the second hypothesis, the lexico-grammatical properties of the target language (TL) and the register features in the TL may be overused in the TT. In both cases, the effect contributes to making translated texts different from comparable texts (i.e. non translated texts in the same register) in the target language. The features selected for analysis are: - agency (frequency of middle vs effective processes) - theme (frequency of marked vs unmarked themes) - grammatical metaphor - transitivity (frequency of relational processes compared to other processes, in particular material processes) - passive voice (frequency of passive voice vs. active voice and passive alternatives) - postmodification and premodification of nominal groups, and density of modification (dense modification compared to full relative clauses)
The hypotheses are further refined for each of the features to be tested and according to the expected discrepancy in the frequency of occurrence of such features as predicted by the typological description of English and German. In general the theoretical argument and linguistic information supporting the hypotheses is sound. However, the means for testing the hypothesis concerning grammatical metaphor (frequency counts of word classes involved in grammatical metaphor: conjunctions, verbs, nouns, and prepositions) are rather weak and therefore any conclusions in this regard will be, at best, tentative, as the author herself will acknowledge in the conclusion.
The hypotheses are tested on a corpus of scientific writing consisting of four subcorpora: German originals, English originals, German translations, and English translations. Each subcorpus contains around 10,000 words. Since they are not exactly of equal size the results (frequencies) are normed on the basis of 10,000 words. They are submitted to statistical tests (chi square). The corpus is annotated with extra linguistic information (in XML) and PoS tagged using the TnT tagger. PoS based queries are used for retrieving information on word classes, passives, predicative and attributive adjectives, premodification and postmodification. The IMS Corpus Workbench was used for this purpose and for retrieving parallel concordances (the alignment tool used was Déjà Vu). More abstract, functional-grammatical and semantic features such as transitivity, agency and theme were coded (manually) and retrieved using CODER.
The analysis is carried out in three stages, first all the results for German and English originals are presented. This allows the author to confirm predictions regarding register features in both languages. This information will also be used later to establish whether a certain feature is actually overused in the translated texts. Secondly, the results for each set of features in the monolingually comparable subcorpora (for example, English originals and English translations) are discussed. Finally, English translations are compared with German originals and German translations with English originals. The analysis of the parallel texts is done on the basis of overall normed frequencies. Only at the interpreting stage are parallel concordances used and then only to elicit extra information to explain ambiguous results.
In the last section the results are summarised and interpreted. Although this way of proceeding is useful in that it highlights the relations between the subcorpora, sometimes it is difficult to keep track of all the results for each particular feature since this requires going back and forward across different sections. (I should note here that there are a couple of inconsistencies across the tables that can only be typos). In any case, the summary of the results is very clear.
Although the hypotheses seemed to be competing with each other, the results show that both TL normalisation and SL shining through can co-occur, affecting different linguistic features and translation directions. One of the possible explanations for some of the results of this study makes an interesting hypothesis to be tested further:
Given a basic commonality in a particular grammatical system, if the TL has more delicate options than the SL, it can afford to let the SL interfere, and where the same compensatory means is used frequently, we get SL shining through. If the TL has less delicate options, it has to compensate, and where the same compensatory means is used frequently, we get TL normalization(p. 223).
Teich insists on a distinction between compensation and normalisation. She argues that one of the advantages of this methodology over others generally used in translation studies is that by requiring the contrastive analysis of comparable original texts prior to the analysis of translations, it can establish with certainty when in the translations there is significant difference in the use of a feature compared to both source texts and target language originals and thus make definitive claims regarding normalization.
Overall, more effects were found in German translations than in English translations, and there was more SL shining through in German translations compared to English translations. In English translations, there were stronger normalisation effects. This leads to the conclusion that German translations are more different from comparable German original texts than English translations from comparable English original texts.
Chapter six provides a summary of the previous chapters and the results, an assessment of the methodology and suggestions of other contexts for applying the model of cross-linguistic variation and issues for further research. Although the interpretation of the results in the light of the proposed model gives enough food for thought and is already very interesting, it would have been interesting to also compare with results in other corpus-based studies of translations; for example with the results obtained by Øverås (1998) in her study of "explicitation", where it was found that there were more effects (more implicitating and explicitating shifts) in Norwegian translations than in English translations. This could bring up the issue of the possible influence of different translational norms for different languages, which is something that Teich does not touch upon. Kenny's study of normalisation (Kenny 2001) would also provide interesting ground for comparison. Kenny found, for instance, that some types of normalisation were more common than others and suggested that how individual translators approach their task may be one of the reasons for diversity of results (Ibid. 188 and 208). This brings me to another issue which Teich does not mention in her study and that is the variation across the individual texts in the corpus. Are the effects stable across all the texts in a given subcorpus?
Apart from that, the conclusions are sound and very interesting, opening up new paths for research in the area. To my knowledge, this is one of the most systematic and comprehensive studies of translations and comparable texts carried out so far. It has a robust theoretical basis, which many other studies in the area lack, followed by a thorough analysis of the results. It also uses more sophisticated tools and methods of analysis than those usually encountered in the area of contrastive typology and translation studies.
REFERENCES Baker, Mona (1995) 'Corpora in Translation Studies: An Overview and Some Suggestions for Future Research', in Target 7:2 223-243.
Baker, Mona (1996a) 'Corpus-based translation studies: the challenges that lie ahead', in Harold Somers (ed.) Terminology, LSP and Translation: Studies in language engineering in honour of Juan C. Sager, Amsterdam & Philadelphia: John Benjamins Publishing Company; 175-186.
Biber, Douglas (1988) Variation Across Speech and Writing, Cambridge, Cambridge University Press
Biber, Douglas (1995) Dimensions of Register Variation: A Cross-Linguistic Comparison, Cambridge, Cambridge University Press
Doherty, Monika (1991) Informationelle Holzwege, LiLi. Zeitschrift für Literaturwissenschaft und Linguistik 84:30- 49.
Doherty, Monika (1993) Parametrisierte Perspektive, Zeitschrift für Sprachwissenschaft 12 (1):3-38.
Hawkins, John A (1986) A Comparative Typology of English and Germn, Croon Helm, London and Sydney.
Kenny, Dorothy (2001) Lexis and Creativity in Translation: A Corpus-based Study. Manchester, St. Jerome Publishing.
Øverås, Linn (1998) 'In Search of the Third Code: An Investigation of Norms in Literary Translation', META, XLIII, 4, 571-588.
ABOUT THE REVIEWER:
ABOUT THE REVIEWER Gabriela Saldanha holds an MPhil in Translation Studies from UMIST, UK and is currently doing a PhD in the same area at Dublin City University, Ireland, where she also lectures on Corpus Linguistics and Translation Technology. Her research interests include Corpus Linguistics, Corpus- based Translation Studies, Translation Technology, Stylistics.