Quantitative Analysis of Correspondence between So and Dus

Correspondents

The overall correspondence between so and dus differs depending on the translation direction (Table 8). In total 148 tokens of so are rendered in Dutch with dus, which amounts to nearly 46% of source text tokens of so and slightly over 44% of target text tokens of dus. These numbers are even considerably higher, with 59.7% and 60.9% respectively, when zero correspondences are ignored. Clearly, when deciding on a Dutch correspondent for so, translators opt for dus in a majority of cases. In absolute terms the numbers are similar when Dutch source texts are translated into English, with 152 tokens of dus translated into English, but in relative terms the differences are more outspoken: almost 41% of target text tokens of so are translations of dus, which compares neatly with the number for the English-Dutch

Table 8 Overall top three of correspondents of so and dus in translations and backtranslations in absolute numbers (N) and adjusted percentages (excluding tokens with zero correspondent)

translation direction, but only a mere 17% of source text tokens of dus are translated with so. Again these rates are somewhat higher if zero correspondences are not taken into account (48.6% and 22.7%, respectively).

The absence of a correspondent for about one quarter of tokens - 23.7% for so and 25.2% for dus - is in line with previous findings on the translation of pragmatic markers and causal connectors (e.g. Bazzanella and Morra 2000; Aijmer and Altenberg 2002; Altenberg 2007, 2010), and can be explained by their syntactic and semantic optionality. Interestingly, though, 27.0% of target text tokens of dus do not have a correspondent in the English source texts and have, therefore, been added by the translator. This compares to a much lower rate of 15.9% of target text tokens of so.

The preference for therefore over so may be explained by two main factors. First, many of the texts included in the DPC have been taken from a fairly formal context. As therefore is considered more formal than so and is certainly more typical of written language (Biber et al. 1999: 887), translators may have felt it more often appropriate to translate dus with therefore than with so.

Second, the preferred position of dus more closely resembles that of therefore than that of so, which may have led translators more easily to the former than to the latter. In over two thirds of cases (68.3%, N = 612) dus occurs in mid-position in Dutch source texts, compared to only 27.9% of times (N = 250) in clause-initial position and 3.8% (N = 34) in final position. Of the 278 tokens of dus that have been translated with therefore 225, amounting to 80.9%, take mid-position in the source text.