New research methods (empirical methods)

On the way from a normative to a descriptive science, the adaptation of empirical methods is an important step forward in translation research.

To be able to formulate truly generalisable statements about the process of translation, the traditional method of comparing source and target language texts can not be regarded as satisfactory any more, even if there has been considerable development in the criteria applied in the past few decades. What is needed now is that researchers should conduct studies under "laboratory conditions", perform­

ing pre-planned and replicable experiments. Translated literature can no longer be regarded as a single data base from a single spontaneous experiment that is there just waiting to be analysed by the researcher.

The application of empirical methods in translation theory cannot be regarded as the result of internal development. James Holmes said in 1970 that researchers of translation theory all obtained their training in other fields. Researchers came not only from the field of linguistics and literature, but also from more distant areas, such as information theory, logic, mathematics, and all of them brought with them the paradigms, quasi-paradigms, models and methodologies of their original discipline. Many turned to translation research attempting to find new areas for research, where they can experiment with new methods. The science of translation has benefited from these imported methods, which have helped to transform it from a normative into a descriptive science (Toury 1991).

Translation research - be it theoretical or empirical - may move in three direc­

tions, depending on whether it is the product, the process, or the function that is in the centre of attention. In these three branches of research, naturally, the possi­

bilities of empirical research also vary. Research on the product or function of translation investigates the reactions of the consumer, that is, the reactions of the reader. Therefore in these cases it is the reader who is the subject of study.

Process-oriented empirical research, on the other hand, studies the process of translating independently of the final version. In such cases the subject of the experiment is the translator.

100

7. New trends in translation theory at the turn of the century

Gideon Toury in his article Experimentation in Translation Studies: Achievements, Prospects and Some Pitfalls (1991) gave a detailed overview of two product-oriented empirical research methods, used to measure readability, clozetests (Snell-Hornby 1983, Puurtinen 1989) and questionnaires (Tirkkonen-Condit 1986).

7.6.1. The cloze test

Cloze tests, as is well known, are very effective in measuring language proficiency in general and the reading skill in particular. Cloze tests can be prepared by delet­

ing every nth (e.g., fifth, sixth, etc.) word from a continuous stretch of text, and the reader has to fill in the missing words. It has also been realised that cloze tests do not only measure how well students understand texts, but they can also show how understandable the text is. If we start out from the assumption that good translation is easier to understand than bad translation and it is easier to extract information from it, then cloze tests may be capable of measuring the readability of translated texts. Nida and Taber have mentioned the use of cloze techniques as early as in 1969: "Actually the only linguistically sound test of ease of comprehen­

sion is the Cloze Technique, which is based on the principle of translational probalilities” (1969:169-170).

The difficulty of such experiments lies in the fact that the readability (compre­

hensibility) of a target-language text produced as a result of translation cannot be assessed independently, only in comparison or relative to other texts, which might either be the original source-language text or other target-language texts. If we choose the first option, claiming that the comprehensibility of a translation should equal that of the original text, then testing can only be done with two native speak­

er groups, speaking different native languages. It is maybe more beneficial if we compare the different translations of the same text with die help of a cloze test.

This is what Tiina Puurtinen did when comparing two translations of The Wizard of Oz. She considered one of the translations, containing difficult syntactic struc­

tures, more difficult to read, and the results of the cloze test confirmed her hypothe­

sis (Puurtinen 1992).

7.6.2. The questionnaire method

Another method to measure readers’ reactions is the use of questionnaires, in which readers, having gone through the translation(s), write down their intuitive impressions about the text(s) read.

In the following, we shall present some of our own questionnaire surveys. The first one was conducted in 1978 where participants had to judge whether a text was an original Hungarian text or a translation as a part of an experiment to demonstrate the phenomenon of so called "quasi-correctness” in translated Hun­

garian texts. (Klaudy 1981 abc, 1984, 1987).The other experiment was conducted in 1993, to explore, on the one hand, what explicitation strategies translators had used in two different Hungarian translations of the same source text (Klaudy 1993), and on the other hand, how explicitation affects readers’ perceptions of the translations.

101

7.6.3. Quasi-correctness and the testing of reader perceptions

In our research in the 1970s we called the Hungarian texts produced via transla­

tion "quasi-correct” (following Ferenc Papp’s term from 1972, 1984) in order to avoid negative labelling which leads to regarding such texts as in some way spoilt or imperfect texts. Instead we wanted to treat translated texts as text type which is worth studying in its own right (Klaudy 1981abc). This same idea, following Gideon Toury’s descriptive approach and Mona Baker’s corpus investigations has now became so widely accepted in translation studies that even the term "quasi- correct" is coming to be regarded as derogatory. The term was also used by Inker!

Vehmas-Lehto, who wrote a book with a similar title comparing Finnish texts translated from Russian with original Finnish texts (Vehmas-Lehto 1989).

As discussed in the chapter on text linguistics, quasi-correctness refers to the difference between whole target-language texts produced as a result of translation and authentic (original) target-language texts. Here we are talking about very sub­

tle differences that are imperceptible on the sentence level: it is the whole of the translated text that differs from original target-language texts. Readers sense some light oddity about the text but cannot identify its source. One of the reasons for this oddity is that the cohesive devices of the source language do not always work in the target language. Another reason for the perceived oddity may lie in slight shifts of stress, and slight distortions in the functional perspective of sentences, which the reader does not even notice if it happens in one or two sentences, but if it happens more often, then it may contribute to the feeling of strangeness.

In our survey, we sought answer to the question whether readers can perceive this strangeness in the translated texts, and whether they are capable of distinguish­

ing between translated texts and original Hungarian texts and then explaining the reasons for the strangeness, etc. The texts used were social science texts, and the readers were experienced readers of such texts, university teachers and students.

Each participant received five texts on different social science topics, out of which the first, the second, and the fifth were translated Hungarian texts, and the third and fourth were original Hungarian texts. They were asked to read the texts, at their speed, neither more closely or more superficially than usual, but exactly in the same way as they generally read. After each text they had to answer three questions:

(1) Do you think the above text is an original text or a translation? (decision) (2) How do you know? (explanation)

(3) If you believe it is a translation, how would you rate it? (evaluation)

The test papers were filled out by 650 participants, and as everybody answered 5 test papers, we evaluated altogether 3250 answer sheets. We cannot go into a detailed description of the results here (for the details see Klaudy 1987), only pres­

ent some of the most interesting findings. The results of the decision and the evalu­

ation of its quality mostly coincided with our preliminary assumptions: the partic­

ipants distinguished separated original Hungarian texts from translations, i.e.

from quasi-correct texts with a reliability of 70%. The divergent reasons given for their decisions, however, showed that the perception of strangeness is to a great extent intuitive. Let us cite just a few examples to illustrate the contradictory reasons:

102

7. New trends in translation theory at the turn of the century

(1) a) It is a translation, because it contains many foreign words.

b) It is original, because it contains many foreign words, which a translator would have translated.

(2) a) It is a translation, because it contains many complicated, complex sen­

tences.

b) It is original, because it contains many complicated, complex sentences, and a translator would have cut them up into shorter ones.

(3) a) It is a translation, because it contains many leftward branching attribu­

tive structures, which are not typical in Hungarian.

b) It is original, because it contains many leftward branching attributive structures, which are only possible to create in Hungarian.

(4) a) It is a translation, because it contains many clumsy, unstructured sen­

tences.

b) It is original, because it contains many clumsy, unstructured sentences, which a translator would have structured better.

On the basis of such contradictory reader reactions, some might say that transla­

tors have a very hard task when trying to meet reader expectations. It is even more interesting that as the findings show the intuitive evaluation of readers worked very well despite the contradictory nature of reader responses. In other words, even if they cannot formulate rules for what they really expect from a translator, they are able to tell whether a particular text meets their expectations or not.

7.6.4. The testing of explicitation strategies

Explicitation is the technique of making explicit in the target text information that is implicit in the source text. Explicitation (implicitation) strategies are generally discussed together with addition and omission strategies in the literature on trans­

lation theory, even though the notion of explicitation is broader than addition or insertion, since it is possible to make something clearer in translation without the actual insertion of additional elements (for more details see III. 3.5.).

In our experiment we intended to find out how one of the explicitation strate­

gies, namely addition, affects the readability of the text. We distinguished between two types of additions:

(1) Obligatory additions, which are necessary to create grammatically well formed target-language sentences. They occur due to missing categories (e.g., article or grammatical gender may exist in one language and not in another). These additions are always carried out by translators, because otherwise they do not get grammatically correct target-language sentences.

(2) Optional additions, which are needed to create a unified coherent target- language text. These may be textual additions, necessary because of the dif­

ferent discourse conventions of languages; or pragmatic additions required because of differences in the background knowledge possessed by

source-and target-language readers. These additions are not always carried out by translators, because they may get grammatically well formed target-lan­

guage sentences even without such additions, the only thing they risk is that equivalence will not be created on the textual level.

Our initial hypothesis was that if readers are to evaluate the quality of translations and choose the translation they like most, then those translations will be rated the most positively which are easier to read, due to being more explicit, containing many additions, insertions and explanations.

The participants in the experiment were 100 second-year English and German majors and teacher trainees from the Faculty of Humanities, University of Miskolc.

Their task was to evaluate two different translations of the same source-language text.

We were lucky in selecting our texts, because it is rare for two Hungarian translations of the same text to be published at the same time. This happened in 1993 in the case of the so-called “Yeltsin dossier" (“Jelcin-dosszié”) (1) A “Jelcin­

dosszié”. Szovjet dokumentumok 1956-ról. Budapest: Századvég; (2) Hiányzó lapok 1956 történetéből. Dokumentumok a volt SzKP KB levéltárából. Budapest: Móra).

The 1956 documents brought back from Russia aroused such an enormous inter­

est among Hungarian readers that two publishing houses, Móra and Századvég both published their translations at the same time.

Both translations were made by professional translators, and both publishers carefully edited the texts, so neither of them contained mistakes or mistranslations.

The contents of the two volumes were not exactly the same, and comparing the translations that appeared in both volumes, the most striking difference between them was that the translations published by Századvég (henceforward “S”) were considerably longer, than those published by Móra (“M"). In the “S” translations the following additions could be found:

(1) addition of linking devices at the beginning of clauses, (2) addition of linking devices at the beginning of sentences, (3) addition of emphasisers,

(4) explanatory translation of toponyms (names of streets, squares), (5) spelling out of abbreviations,

(6) explanatory translation of military terms,

(7) apparently unjustifiable additions which might only be explained by the individual preferences of translators.

In the first part of the questionnaire (global reading), participants were asked to read the two translations and to decide whether they perceived any differences between the two versions, and if so, which translation they considered to be bet­

ter. They had 20 minutes to do this task. In the second part (text analysis), they had to find additions in both translations. They had 25 minutes to complete this task.

The answers given to the first part of the questionnaire confirmed our hypoth­

esis, according to which readers will prefer more explicit translations. On the basis of global reading, that is, intuitive judgments, “S” translation (420 words passage

7. New trends in translation theory at the turn of the century

chosen randomly) was given a higher rating by 70% of the participants than the

“M” translation (375 words passage chosen randomly).

The second part of the experiment, hovewer, brought surprising results. After the actual analysis of the texts, i.e. after the identification of additions, several readers modified their opinions. The analysis raised the question whether it is cor­

rect to provide easy-to-read, fluent translations in the case of military documents.

Many claimed that the less reader-friendly version was a better reflection of the atmosphere of the times.

This questionnaire-based survey of reader reactions showed that explicitation may have favourable effect on readability. At the same time, however, it raises the question, whether it is desirable to consciously aim at ease of reading, because attempts to enhance readability may risk authenticity (Klaudy 1993b).

7.6.5. The introspective method

The introspective method is typically one that has come into translation stud­

ies from other disciplines. In psychology, it was used as a research tool by the Würzburg school for the experimental investigation of thinking processes as early as in the first decades of our century. Thinking aloud was used by Claparède in the 1930s to explore what hypothesis-making processes take place in the minds of the participants during the process of problem solving (more details in Lörschner 1991). In the 1970s, data obtained from introspection were used to investigate language acquisition strategies (Cohen 1984).

To investigate thinking processes during translation introspection was first used by Hans Krings (1986). He asked language learners to speak into a tape recorder and tell everything that comes into their minds while translating. The recordings were transcribed and these think aloud protocols were analysed according to vari­

ous criteria.

Königs’ experiments (1987) aimed at determining the proportion of automatic and non-automatic processes. The analysis showed that in translation, thinking takes place in two blocs: the automatic bloc (Adhoc Block) contains the use of the translator’s internal vocabulary and previous experience, whereas the non-auto­

matic bloc (Rest Block) contains the conscious linguistic and stylistic decisions, the adjusting of the text to the aim of the translation and the needs of the audi­

ence, and taking into consideration information regarding the author of the text, etc. Borsch (1986) pointed out that for professional translators many processes are likely to be highly automatised and therefore not accessible via verbal report procedures (for more details on TAP research see Jääskeläinen 1998).

In document TRANSLATION LANGUAGES (Pldal 102-107)