20.11.2014 Views

Translation Universals.pdf - ymerleksi - home

Translation Universals.pdf - ymerleksi - home

Translation Universals.pdf - ymerleksi - home

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

110 Jarmo Harri Jantunen<br />

Figure 2. The span. “L” strands for the left and “R” for the right side, the numbers<br />

mark the distance from the node. (An approximate translation in brackets.)<br />

context. This Keyword in Context (KWIC) analysis is utilised to extract the<br />

immediate colligates and collocates of node words. After the relevant concordance<br />

lines (i.e. lines which include the node) are extracted from the corpus,<br />

they are sorted manually according to the word class in a given position. The<br />

span is limited to the space of two words to the left and two words to the right<br />

of the node. However, not only the words are counted but also the clause beginnings<br />

or ends, which could turn to be distinctive parameters in the analysis<br />

(see Figure 2).<br />

3.2 Statistical procedures employed to analyse the similarity and difference<br />

On the grounds that “one can never be entirely sure that the observed differences<br />

between two groups of data have not arisen by chance due to the inherent<br />

variability in the data” (Oakes 1998:1), I have adopted a number of statistical<br />

procedures to avoid misconstructions of the data. The chi-square (x 2 )test 7<br />

(Butler 1985; Oakes 1998) is used to test the significance of observed frequencies<br />

in different subcorpora. Furthermore, statistical methods are used to test<br />

the significance and strength of collocations. To measure the significance, there<br />

are several tests available, of which z-andt-scores and Mutual Information (I)<br />

are the most commonly used (Barnbrook 1996:94–100; for the range of tests,<br />

see also Oakes 1998).<br />

According to Barnbrook, it can be difficult or even impossible to select<br />

one test that best evaluates the significance of the collocation in question<br />

(ibid. 101). This view is shared by Stubbs (1995), who claims that tests can be<br />

confusing and they must be interpreted with care. Both Stubbs and Barnbrook<br />

suggest that to achieve a balance between different tests it is probably better to<br />

use more than one statistical measure. In his analysis, Barnbrook (ibid. 100–<br />

101) reports that the three tests mentioned above provide different kinds of<br />

information on the significance of collocations: while both the z-score and the<br />

Mutual Information measures underline the significance of low frequency co-

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!