Levels of explanation and translation expertise

Abstract

The observable activity of translation, the series of text comprehension and text production bursts we identify as translation, is the result of the activation of complex underlying cognitive systems. In the conduct of research it is often useful to approach such complex systems using a ‘levels of explanation’ framework. This paper considers David Marr’s (1982) three levels of explanation as they might apply to understanding translation and translation expertise more robustly. In cognitive translation studies to date, we have not really extended our understanding of expertise much past the second (algorithmic/representational) level in Marr’s scheme; we have failed to grapple as effectively as we might with the problem of how the second generation computationalist expertise constructs we adopted almost twenty years ago could be integrated with, for instance, connectionist neural network models of the mind, creating a third generation of expertise models. This paper offers some frameworks laying out how that end might be achieved using, for instance, symbolic connectionism and implementational connectivism. Further, it argues that given the overtly symbolic nature of translation language processing, cognitively-oriented translation scholars are uniquely suited to benefit from approaches that bridge the divide between symbol processing models and connectionist ones.