Newson, D. (1998). Translation and foreign language learning. In K.Malmkjær (Ed.). Translation and language teaching: Language teaching and translatio

Newson, D. (1998). Translation and foreign language learning. In K.Malmkjær (Ed.). Translation and language teaching: Language teaching and translation (pp. 63-68). Manchester: St. Jerome Publishing.


Newson (1998) proposes a simple model in teaching translation, especially from the mother tongue, in an EFL context. The study involved German learners of English. Translation is used as a test of proficiency in Germany.

The author argues that there are alternatives to sentence level translation, and that it is possible to include true translation in the language classroom. True translation refers to translation that is done for a specific purpose and audience in mind.

Teaching translation in the classroom is beneficial on two levels. As a short-term benefit, doing so could prepare students for the mandatory translation exam during their finals; and as a long-term benefit, students could gain insights into the workings of a language.

Newson (1998) claim that translation should not be used as a teaching and testing tool because:
1. It encourages thinking in one language and transferrence into another, with accompanying interference.
2. It deprives teacher and learner of the opportunity to benefit from the accruing advantages of working within one language.
3. It gives false credence to the naive view that there is such a thing as simple word-to-word equivalence between languages.
4. It does not allow or make easy the achievement of such generally accepted foreign language teaching aims as:
a) Emphasis on initial fluency in spoken language
b) Attention to the controlled introduction of selected and graded structures (60s style) or communicative competence strategies (90s style)
c) Attention to controlled introduction of and mastery of selected and graded lexical items
d) The use of situationalized, contextualized language
e) communicative language use
f) Learner-centered language learning
g) There is no observable learning effect, either of new vocabulary or structural items. This is not surprising since each translation task provides normal(l)y only one (random) example of new language items; there is no repetition and practice as in classic forms of language learning and teaching, no grading and no stru(c)turing (pp. 63-64).

Here, Newson (1998) presents several ungrounded claims that could easily be refuted.

Newson (1998) criticized the use of translation as a text of language competence because it is not a measure of the target language since they only present random translation problems and the chosen text presents an unpredictable sample of the target language, making it impossible for the diligent learner to prepare for systematically and reliably and for the teacher to supply planned and graded practice.

A possible solution to this dilemma is to retrieve authentic representative language from a representative text data bank. To do this, the instructors needs to compile translation texts that adhere to linguistic criteria. Specifically, these should be non-literary, originall written in the source language, about English-speaking countries, and of a specific length. This would make the translation task more practical and more predictable.

The second step is to sort for word frequency. That way students will only need to translate and become more familiar with the most frequent meaning of words. The readability measure of a text should also be probed to determine its difficulty. This is done by counting the number of passive constructions in the text and measuring sentence length. Both frequency count and readability measurement allows for the grading of the texts according to their difficulty.

From here on, the instructor could create a data bank that includes representative examples and not just isolated and chance occurrence of target constructions and items.

Finally, systactical contrastive studies can also be conducted to predict likely translation or interference problems. These steps are more systematic and generalizing than 'false friends.'

Building such a model would help make translation practice more contextualized and not just inauthentic translationspeak generated for the sole purpose of illustrating a point.