European Society for Translation Studies

How to choose the most relevant parameters to be controlled in an empirical study

How to choose the most relevant parameters to be controlled in an empirical study


Daniel Gile
12 August 2006

In a previous text (Sampling 2: Generalizing from case studies), it was explained that in order to make findings of studies as generalizable as possible, researchers, who cannot attend to all relevant aspects of the phenomenon under investigation in any single study, have to rely on replications and attempt to attend to the parameters and parameter values which they believe are the most relevant; “most relevant” in this case means most likely to have a non-negligible effect on the dependent variable studied. But how they know which are the most relevant parameters and parameter values?

Obviously relevant parameters are those that have been previously shown to affect the dependent variable. For instance, if previous research has shown that formally trained translators have certain working methods which affect certain quality parameters, in a study on the effect of experience on translation quality, the formally trained/ self-trained status of subjects is relevant and needs to be incorporated in the design, for instance by including in the samples the same proportion of formally trained and self-trained translators.
Another resource for the selection of relevant parameters to attend to is theory. A theory is an inter-related set of assumptions on the nature, make-up and/or operation of an entity and these take on board factors considered most relevant. A linguistic theory of translation is essentially built with linguistic factors. A cognitive theory of translation also includes other factors, linked to human cognition. A more culture-oriented theory of translation will point to cultural factors, ideological factors, etc.

The selection of relevant parameters on the basis of a theory will be biased towards those factors which the theory considers most important, at the risk of neglecting other factors and thus designing studies which will not be generalizable – and sometimes not even valid (a study is valid if it studies what it is supposed to study; for instance, studies which neglect the fact that translation operates on texts in a context with a communication intention and which measure what translators do with single words without a context or a communication intention may be measuring action quite remote from translation).

It is therefore important to also rely on another important source of assumptions on the relevance of various factors and on the relevant values of the associated parameters, namely direct, experiential knowledge of the environment under study. In the field of TS, in particular, researchers must realize that professional translators, who are in daily intimate contact with their trade, are likely to be sensitive to factors that theories have not taken on board properly yet. While some of their beliefs may turn out to be unjustified, they are probably a better starting point for further investigation than linguistic, cognitive, sociological and other theories as such. Ideally, such theories should integrate experiential input as a starting point, test it and then decide to keep it or discard it.