Problems of phonemicization

Irish short vowels revisited

September 1, 2016

9th Celtic Linguistics Conference, Cardiff University, Cardiff, Wales

 download handout

The file is open to annotations via Hypothes.is. Comments are always welcome!

Categories:  Irish Phonetics-phonology interface

(with Máire Ní Chiosáin)

 Share on Twitter

We address a long-standing issue in the phonology of Modern Irish: the phonological status of the distinction between front and back short vowels. We argue that the approach inherent in most theoretical work on the topic, which privileges traditional ‘phonemic’ criteria such as predictability of distribution, misses important generalizations and cannot be sustained. Instead, we suggest that front and back (non-low) vowels in Irish are best thought of as phonologically ‘distinctive’, but not (significantly) ‘contrastive’ units.

We reconsider the arguments for the noncontrastive status of the front/back distinction in short vowels in Modern Irish, on the basis of a cross-dialectal acoustic study. Our results show that the distributional generalizations offered in Roibeard Ó Maolalaigh’s (1997) summary of the traditional description literature, which suggests the distinction is largely allophonic, are by and large correct. Our quantitative analysis, which relies on model comparison methods applied to generalized additive mixed models, also shows that front and back vowels form distinct categories in the output of the phonology: models that assume more than one backness category per height achieve better fit than ones assuming a surface ‘vertical’ system.

We conclude that front and back vowels form distinct categories that are accessible to the phonological grammar, and offer an interpretation of the facts centred around the notion of ‘quasi-phonemes’: phonologically distinct elements that do not stand in contrastive distribution. We argue that this framework allows us to unify the description of the low functional load of the distribution and the analysis of vowel backness alternations in the synchronic grammar. We show that our analysis, which does not predict fully complementary distribution, is superior to previous approaches that relied on ‘vowel separation’ rules to achieve the same outcome, since it correctly predicts the possibility of exceptions.

This work builds on that presented at OCP13 and FiSK 2016.



About me

I’m Pavel Iosad, and I’m a Senior Lecturer in the department of Linguistics and English Language at the University of Edinburgh. ¶ You can always go to the start page to learn more.

elsewhere

Updates

Subscribe to the  RSS feed, or follow me on Twitter at  @anghyflawn.

Search