Sources of phonological variation in a large database for Dutch dialects

The so-called Goeman-Taeldeman-Van Reenen Project (GTRP) consists of a large online database of 613 local dialects of Dutch on the basis of which the phonologies of these dialects can be systematically compared. In this paper we present a quantitative investigation of an aspect of the reliability of the GTRP data. To this end, we performed a series of statistical analyses in order (1) to detect to which extent the various investigators involved affected the data regarding the palatalisation and velarisation of coronal nasal-plosive clusters, and (2) to determine the weight of the effects exerted on the variability in the data by the investigators, dialect geography, diachronic and synchronic internal factors as well as several parameters of usage. The findings are relevant for both theoretical and methodological reasons. In the former connection, we discuss the effects of the phonological as well as frequency factors determining palatalisation and velarisation in this environment. In the latter connection, we make suggestions for the calibration and enrichment of phonological databases.