Didactics of Translation: Text in Context

Testing achievement in translating and evaluating translations

According to Farahzad (1992: 271), evaluating and scoring students’ work should be done following “concrete criteria”, and not on the basis of a subjective approach. Farahzad further adds that “our classes are large and we must set up standards for assessment which are accepted as tolerably objective, both by our students and ourselves as teachers” (ibid). For this goal, House’s model of translation quality assessment “as an empirically based attempt to make the evaluation of translations more objective”, seems the most appropriate because “instead of taking the psychological category of readers’ intuitions and beliefs as the cornerstone of translation evaluation, the model looks at language in social life and focuses on texts, the products of human decision processes that are the most tangible and least ambiguously analysable entities (Ibid: 118).
For the construction of a valid translation test, Farahzad (1992: 272) suggests the use of both limited-response and free-response techniques. The first will enable teachers to deal with certain delicate points identified by contrastive analyses between the source and target languages such as: interference of the source language syntax, the lack of equivalents of certain language patterns in the target language, the mismatches of lexical items between source and target language, etc. The second will help teachers decide on the degree to which the students “can treat a text as a piece of discourse and translate it adequately” (Ibid).

1. 
   Limited-response tests:
   

In this kind of test, there are three types of questions: error recognition, multiple choice questions and the provision of target language equivalents. (Ibid: 273). In error recognition, the examinees have to identify errors in translated sentences or texts and then correct them. They, furthermore, have to justify their answers. The errors could be related to meaning, grammar, word choice, punctuation, transliteration or individual text function. In multiple-choice questions, the examinees have to choose between two target translations of a SL sentence or text by encircling the better one. Finally, in the provision of target language equivalents, the students are asked to give target language equivalents for certain SL words and phrases.

2. 
   Controlled free-response tests:
   

In this type of test, the testees’ translations are controlled by the source language text, i.e. their translations seek to emulate the ST’s content and sometimes even its form. On the other hand, these tests allow the testees the freedom to decide on certain equivalents rather than others, to opt for certain arrangements or adjustments or to leave out certain source text items completely. According to Farahzad (1992: 274 – 275), this type of test is:
an integrated test examining several components of translation at a time, such as comprehension of the ST, accuracy in terms of content, appropriateness of grammatical forms, and choice of words, etc.
Farahzad further adds that this procedure encourages testees to deal with the source text as a “coherent whole rather than as a string of disconnected sentences” (Ibid). However, to obtain valid results when assessing this test, Farahzad puts forward some conditions which translation teachers have to observe. One of these is adapting the degree of difficulty of the test to the level of the testees:
The test must match the examinees’ level of command of both source and target languages as well as their level of translational competence (Ibid).
It has to be noted, however, that the controlled free-response test could also be used to measure the testees’ degree of textual awareness. Following this type of test, the testees could be asked to provide justifications or arguments for their translation decisions using insights from linguistic-based models such as Hatim & Mason (1990) and House (1997). Similarly, the limited-response test could also include questions asking the testees to justify their choices by invoking, for example, House’s criteria for translation quality assessment: semantic and pragmatic equivalence, in addition to equivalence of the individual target text function with that of the source text.
Both limited-response tests and controlled free-response tests include the scoring of the testees’ work as a final stage in the evaluation of their achievement. For Farahzad (1992: 276), when scoring, the examiner has to consider two criteria: the unit of translation and the weight to be given to various aspects of this unit.
The unit of translation could be either the sentence or the text. If it is the sentence, the examiner should examine two features, namely accuracy and appropriateness and should give a score to both. Accuracy obtains when “the translation conveys the information in the source text precisely, without addition or deletion”, whereas appropriateness refers to whether

the sentences “sound fluent and native, and are correct in terms of grammatical forms” (Ibid: 276).

Farahzad proposes the following marking scheme for the evaluation of the sentence as a unit of translation:

x A correct sentence which does not preserve the content receives no score.

x If the target version conveys the message, but in a structure which distorts the meaning, the translation receives no score.
x If the message is conveyed, albeit in a grammatically unnatural form, the translation receives half a score.

As for the error recognition test, Farahzad suggests giving a score when the testee detects the error and a score when he corrects it. Only half a score is given if the testee fails to provide a correction of the error.

If the unit of translation is the text, on the other hand, then scoring, according to Farahzad, could take one of two forms: holistic scoring or objectified scoring (Ibid: 277).
In holistic scoring, the examiner reads the translation once and gives credit for each of the following features: accuracy, appropriateness, naturalness, cohesion and style of discourse (i.e. choice of words and grammatical structures). Farahzad does not favour this kind of scoring because of its subjective character and prefers the use of objectified scoring which is said to be more reliable even though it requires a longer time to complete.
In objectified scoring, the examiner has to read the translation twice: first, to check accuracy and appropriateness and second cohesion and style. Moreover, while considering accuracy and appropriateness is said to take place at the level of the sentence as the unit of translation, examining cohesion and style is said to take place at the textual level. Each of these features (accuracy, appropriateness, cohesion and style) is then given a particular score. This kind of scoring, according to Farahzad, is ideal with large classes.
It may be remarked that although an attempt is made in Farahzad’s work to deal with the nature of the ST and TT in terms of equivalence relations with regard to accuracy, appropriateness, cohesive links and style, the nature of the relationship between the ST features and language users (writer, translator and receiver) is disregarded. Thus, no reference is made to the different dimensions of context (communicative, pragmatic and semiotic). Similarly, no attempt is made to deal with the relationships in a translation between texture, structure and higher contextual

dimensions; i.e. a translated text, which is the result of motivated choices on the part of the translator, is neglected.

As shown above, following House (1977) (1997), to deal with this kind of relationship, the linguistic and situational features of the ST have to be analyzed and the function of the individual text has subsequently to be established, thus leading to the ST’s textual profile.
Farahzad seems to be concentrating on what House calls “overtly erroneous errors”, i.e. errors which result from differences in the denotative meanings of the source and translation text elements, or from a violation of the target language system. In contrast, he disregards “covertly erroneous errors”, i.e. contextual errors and functional matching.
In conclusion, when scoring students’ translations, one should not concentrate just on linguistic equivalence relationships between the ST and TT. One should go beyond that and see the extent to which the student has been successful in interacting and interpreting the ST features and in preserving the individual text’s function as stated in the ST textual profile. While checking accuracy and appropriateness in a translation test is quite suitable for beginning translation students, it is necessary for advanced translation students to demonstrate that the appropriateness of particular items is subject to the position of these items within an overall textual plan involving relations between texture, structure and discourse (Hatim & Mason 1990).
CHAPTER V