Variance and Informativity in Different Measures of Linguistic Acceptability
Thomas Weskott and Gisbert Fanselow
431-439 (complete paper
or proceedings contents
This paper deals with the issue of variability in different measures of linguistic acceptability. It has been argued that acceptability, when measured with the magnitude estimation method (ME), reveals the underlying gradience of linguistic judgments, while other measures, like Likert-scale judgments (e.g. on a 7-point scale), do not provide the same amount of information about the gradient basis of linguistic judgments. This assumption will be questioned by claiming that not only do Likert-scale judgments provide the same amount of information about a given empirical hypothesis, but also that the inherent variability of ME judgments makes them more susceptible to the production of spurious variance. These claims will be backed up by reporting data from a study which compares three different measures of linguistic acceptability: categorical, Likert-scale, and ME. This study was concerned with a phenomenon which has attracted much attention in the literature on linguistic judgments, namely word order variation in German. For the empirical hypothesis the authors investigated, all three measures of acceptability contain the same amount of information relevant to hypothesis testing. In addition, it will be shown that ME judgments contain more spurious variance and hence are more vulnerable to a reduction in statistical power. These results are then discussed by relating them back to the question of informativity and variability of different acceptability measures.
Proceedings of the 27th West Coast Conference on Formal Linguistics
edited by Natasha Abner and Jason Bishop
Table of contents