Iosad, Pavel. 2021. The ATR/Laryngeal connection and emergent features. MS., The University of Edinburgh. Submitted to Bert Botma, Florian Breit, Marc van Oostendorp & Marijn van ’t Veer (eds.), Primitives of phonological structure
The file is open to annotations via Hypothes.is. Comments are always welcome!
You are welcome to download and read this paper. I especially welcome feedback on it. As it is not yet published in final form, if you want to cite the paper, please check with me first. Thanks.
I consider an unusual sound change and synchronic alternation in south-eastern dialects of Welsh, traditionally referred to as ‘provection’ (Welsh calediad). This phenomenon involves the ‘devoicing’ of stops following stressed vowels. I suggest that it may be the result of a typologically rare sort of vowel-consonant featural interaction, where ‘devoicing’ is triggered by tense vowels. This is unexpected typologically, since tongue root advancement — often a correlate of ‘tense’ vowel quality — is more frequently associated with voicing rather than devoicing in obstruents. I argue that in Welsh the phonologization of the quality distinction in stressed vowels has created a system where such an interaction is dispreferred.
I suggest that the unusual consonant-vowel interaction in south-eastern Welsh is facilitated by the phonological ambiguity of the ‘tenseness’ contrast in vowels. Although the distinction between ‘tense’ and ‘lax’ vowel pairs is made consistently in most Welsh dialects, it is often fairly inert phonologically. This makes it possible for it to become involved in a variety of phonological processes. I survey the diversity of ways in which quantity-quality interactions in vowels found across Welsh dialects, and argue that the attested multiple paths of phonologization support a bottom-up, substance-free approach to phonologization. I suggest that the particular strength of the substance-free model lies in its ability to account for different — sometimes drastically so — phonological behaviour emerging from ambiguous data.