Sonority Sequencing in Polish: the Combined Roles of Prior Bias & Experience

Jarosz, Gaja and Amanda Rysling. Oct 2016. Sonority Sequencing in Polish: the Combined Roles of Prior Bias & Experience. Annual Meetings on Phonology 2016, University of Southern California.

Abstract

Recent findings demonstrate sonority projection effects in English, Mandarin, and Korean: speakers exhibit gradient sonority sequencing preferences among illicit initial clusters, and these preferences mirror the SSP (Berent et al. 2007, 2008; Daland et al. 2011; Ren et al. 2010). While some argue that these results support an innate bias (Berent et al. 2007, 2008; Ren et al. 2010), recent modeling studies question this conclusion, showing that certain computational models can detect SSP preferences from lexical statistics in these languages (Daland et al. 2011, Hayes 2011). In a recent response, Jarosz (2015) finds that Polish presents a particularly good test case to differentiate these competing hypotheses: the same phonotactic models that succeeded in the earlier studies do not predict sonority projection based on the lexical statistics of Polish. This follows primarily from the fact that sonority plateaus make up nearly half of all initial clusters. Jarosz also shows that children acquiring Polish nonetheless exhibit sensitivity to the SSP, producing attested onset clusters with larger sonority rises more accurately in spontaneous productions.

The present paper expands on these findings by examining the phonotactic knowledge of adult Polish speakers. We report the results of an online acceptability judgment experiment focusing on initial clusters and present the results of computational simulations evaluating the ability of phonotactic models to predict participants’ ratings. Our main findings are that 1) SSP is predictive of adults’ ratings, 2) sonority projection arises in both attested and unattested clusters, and 3) while phonotactic models have significant preditive value, they do not subsume the SSP preference observed in the participants’ ratings.