Ms. Inga Arffman' s Academic dissertation 14.9.2007

Arffman, I. 2007. The problem of equivalence in translating texts in international reading literacy studies. A text analytic study of three English and Finnish texts used in the PISA 2000 reading test University of Jyväskylä. Institute for Educational Research. Research Reports 21.



The purpose of the study was to examine the special problems of equivalence that arise when translating texts in international reading literacy studies, where equivalence of difficulty between all the different-language reading texts is a key prerequisite for the validity of the entire test. Knowledge of such problems, in turn, helps to increase the quality and the degree of equivalence of the translations used in these studies and, in the end, the validity, fairness and equity of these studies.

The data of the study consisted of three English source texts and their Finnish translations used in the PISA 2000 international reading literacy study. Each of these three texts represented a different text type: one was an expository text, one a narrative text and one a non-continuous text. The method for analysing this data was comparative linguistic text analysis: the English source texts and their Finnish translations were analysed and compared at different linguistic ranks and strata, with the objective of locating, analysing and assessing potential non-equivalences of difficulty in the texts. These, in turn, were determined on the basis of cognitive theories of reading. The analysis was mainly qualitative but also contained quantitative comparisons.

The analysis pointed to six main categories of problems leading to non-equivalences of difficulty between the English and Finnish texts: problems related to languagespecific differences in grammar, problems having to do with language-specific differences in writing systems, problems concerning cultural differences, problems associated with language-specific differences in meaning, problems related to the strategies used and choices made by the translators, and problems linked with editing.

The problems differed considerably between the text types, both in number and quality. In absolute numbers, they were clearly most numerous in the narrative text; when proportioned to the length of the texts, however, they were most frequent in the non-continuous text. In each case, moreover, the fewest problems were found in the expository text. Also the distribution of the problems varied noticeably between the text types. In the expository text, the problems were related to the medical and anatomical terminology of the text, the resulting shift in register in the two texts, differences in word length, and interference. In the narrative text, the principal problems were the numerous polysemes, metaphors and personal pronouns used in the text, the stylistic flattening of the Finnish text, the cultural unfamiliarity of the text for Finnish readers, the translators’ improving and explicating the Finnish text, and in the Finnish text, a loss of interestingness. And in the non-continuous text, problems were mainly caused by compact language, thematic and textual incoherence, and unidiomaticity.

As a result of all the problems, none of the three Finnish translations was fully equivalent in difficulty to its English source text. Most of the individual nonequivalences, however, were small and largely evened out when examined collectively, leaving the true equivalences of difficulty between the English and Finnish texts relatively insignificant.

The results suggest that it will probably never be possible to attain full equivalence of difficulty between all the different-language texts in international reading literacy studies. A relatively high level of equivalence, however, seems to be attainable. To this end, the source texts should referably be easily translatable and hence not contain too many problematic words, structures and topics. While actually translating the texts, the translators could have two parallel source texts in two languages on which to base their translations; they should, moreover, avoid both too literal translation and undue improvement and make wise use of adaptations and compensations. When verifying and judging the quality of the translated texts, statistical item analyses should be complemented with thorough linguistic comparisons, by backtranslations, and by having outsiders read the texts and answer the questions. Throughout the process, qualified translators and editors should be used to make and edit the translations; both also need sufficient time to do their job properly.

Keywords: equivalence, validity, reading, comprehension, difficulty level, translation, assessment