On the Impact of Aptness, Conventionality and Familiarity on Metaphor Processing from a Meta-analytical Point of View

Psycholinguistic research into metaphor processing is burdened with empirical problems as experiments provide diverging evidence on the impact of conventionality, familiarity and aptness, and with conceptual issues as the interpretation and operationalization of the three concepts mentioned, as well as the related predictions which can be drawn from theories of metaphor processing, are controversial in the literature. This paper uses tools of statistical meta-analysis in order to bring us closer to the solution of these problems and reveal future lines of research.

Large-sample confidence intervals of information-theoretic measures in linguistics

This article explores a method of creating confidence bounds for information-theoretic measures in linguistics, such as entropy, Kullback-Leibler Divergence (KLD), and mutual information. We show that a useful measure of uncertainty can be derived from simple statistical principles, namely the asymptotic distribution of the maximum likelihood estimator (MLE) and the delta method. Three case studies from phonology and corpus linguistics are used to demonstrate how to apply it and examine its robustness against common violations of its assumptions in linguistics, such as insufficient sample size and non-independence of data points.

The Words that Count

Computer assisted discourse studies (CADS) undoubtedly offers great prospects in our attempts to observe and understand the use and social effects of language in context, but it does so with a caveat: we need to be constantly aware of our significant (and often implicit) assumptions when attempting to reach beyond electronically analysed masses of text into assessments of pragmatic (and so social) behaviour. This paper aims to remind us of the need for reflection on our most basic assumptions as we begin to make use of more complex and refined procedures and start to make more ambitious claims about what various corpora can show us. It is argued that concepts such as the word, tokens, types and frequency require constant re-evaluation, in particular when we are using data that have been extracted from their original textual (and so contextualized) sources in the creation of corpora. It is hoped that a small contribution can be made to the debate about the empirical approach to understanding language, perhaps in terms of methodologies to be utilized, the potential extent and limits of CADS, or in terms of presenting or interpreting the results and conclusions of published studies.