Investigating Non-Linear Trends in Questionnaire Studies Using Contrastive Bipolar Scales

ABSTRACT

The following study provides evidence that the use of a high number of test items in a subjective rating task with a bipolar scale does not affect the outcome detrimentally, a question hitherto unexplored in quantitative linguistics. By ‘contrastive bipolar scale’ we have in mind a scale with two different, explicit ‘options’ (here words), one at each end. We analyse a large-scale rating study consisting of nearly 1,000 participants and almost 700 pairs of word items per participant. A pair consisted of two synonymous words from a non-standard Silesian variety, one of German origin (a loan), one from the Polish Standard language. The judgement to be given was an estimation of the subjective frequency of the use of the German loan in comparison to its Polish equivalent in colloquial Silesian speech. We use non-linear regression analysis, GAMM, to examine if the presentation order of words affects the mean ratings. Findings show that results are not affected by either the high number of test items or the order of the presentation of word items.

Acknowledgments

We thank the participants of the study and colleagues in Katowice who organized the interviews, and last not least, the anonymous reviewer of the first version of the paper.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. This research project with the title German-Polish lexical parallels in contemporary spoken Silesian is financed by The Federal Government Commissioner for Culture and the Media. It is due to be finished by the end of 2019.

2. When speaking of Silesia we refer exclusively to Upper Silesia. Lower Silesia is not at issue, because by the end of the nineteenth century all former Slavic dialects were already almost entirely extinct.

3. The high number of respondents is necessary to analyse several sub-regions of Upper Silesia. It is not clear at the moment how widespread koineization of Silesian is and to what extent this question concerns loans from German.

4. However, about every tenth respondents did not answer this question in the socio-biographical section of the questionnaire, which may be due to political caution. In other words, German may be somewhat more common among the respondents than these figures suggest.

5. As a matter of fact, one-third of the test items is or was shared by at least one Polish dialect other than Silesian. (Words from German also witnessed in Standard Polish were not considered.) For details and effect of this aspect cf. Hentschel (in press).

6. Cf. Hentschel (in press) for a descriptively more detailed presentation of the three measures of central tendency for items and participants, and for the consequences our statistical analysis may have for a codification of Silesian. The latter is a topic of emotional debate in language policy in Silesia and Poland.

7. The set of tested 689 items contains a couple of items (less than five) that turned out not (or most probably not) to be of German origin. Nor are they mediated by German. They will thus have to be removed from the list in the future analysis. This was only discovered after this analysis had been completed.

Additional information

Funding

This work was supported by the Federal Government Commissioner for Culture and the Media [ZMV I 2 – 2517DK06ZZ];