Our next meeting will be Tuesday March 25th at 10am in room 117. Ray will be presenting his thesis research titled “Moving from Compensatory Lengthening to Partial Assimilation in Chilean Spanish.” This is in part in preparation for his talk at CLA, so there will be a discussion afterwards about the research. Here is the abstract:
Spanish lost the geminates and long vowels originally present in Latin, but the understudied Chilean dialect displays phonological processes that seemingly create prosodically ill-formed outputs, which have been neglected in the literature. The deletion of coda /ɾ/ and word-final /s/ appears to trigger the compensatory lengthening of following onsets. Meanwhile, the deletion of intervocalic voiced stops seems to leave surrounding short vowels in hiatus with no lengthening, which is opaque given that Chilean overwhelmingly resolves hiatuses through glide formation. This paper looks to formally represent these patterns by constraining existing theories of features and prosodic structure, proposing a segment-prosody mapping to account for the findings.
I travelled to Chile and recorded seven pairs of adult native speakers in the Santiago area performing three tasks, and controlled for morphological and phonological environment, as well as stress of each token to look for relevant effects. I find that coda /ɾ/ strictly lengthens sonorants, and word-final lenited /s/ lengthens any following initial consonant (3). Only the former can occur morpheme-internally, while both processes are more likely to occur over smaller morphological boundaries (e.g. host-clitic).
I argue that lenition involves a loss of features, in the case of /s/ to [h] to Ø, where Ø still retains a [+cons] feature in its structure, allowing it to be a target for feature-filling from other consonants, creating the illusion of gemination from what is truly near-total assimilation. Next, I argue that rhotics are underspecified for place and sonorant features, making it a subset of other sonorant consonants, thus, also a target for feature-filling as a phonotactic repair. Finally, intervocalic voiced stops lenite but do not fully delete because their surrounding vowels are kept apart with no hiatus to resolve. The feature [+cons] is also retained here, restricting feature-filling to consonant melodies, so vowel melodies cannot spread into the position.
I conclude that in true CL, all features of the coda are deleted, thereby stranding a mora and allowing vowels (the optimal mora-bearing units) to associate to it. By contrast, when few features are linked to a mora, whether due to underspecification (/ɾ/) or lenition (/s/), features from adjacent segments spread to those positions, but as long as some featural trace is left in the representation, moras are not stranded, and assimilation is constrained to feature-filling.
There will also be a Zoom link open for this meeting at the following link:
https://mcgill.zoom.us/j/84127437214?pwd=3U6LIdUJyJi6TnacFR1OwPRetKNAAN.1
Meeting ID: 841 2743 7214
Passcode: 674595