This is the abstract of a TULCON14 presentation.
Diana Gil Hamel (University of Toronto)
One of the most unique features of the Irish language (as well as its sisters, including Scots Gaelic and Welsh) is the morpho-phonological system of initial consonant mutation. Irish has two principal mutations: lenition, which converts plosives to fricatives while preserving voicing and debuccalizes or deletes radical fricatives; and eclipsis, which replaces voiceless plosives with voiced plosives, voiced plosives with nasals, and fricatives with glides. These two mutation paradigms are robust in modern Irish and both apply to nouns, verbs, and adjectives to encode a variety of grammatical information (compare a cat ‘her cat’ and a gcat ‘his cat,’ which differ only in initial mutation).
This paper concerns the case of English words loaned into Irish and the degree to which speakers incorporate them into these mutation paradigms. Theoretically, we can understand that Irish speakers encountering English loans could incorporate them into their morpho-phonology in one of two ways: either by leaving the English words untouched or by treating them like Irish
words and mutating them as necessary. Previous studies of conversational and elicited data, like Nancy Stenson’s 1990 survey of Ráth Cairn Irish, established that different initial consonants are mutated at different rates in loans, but go no further in attempting to explain why the consonants
differ in this regard. My reading of Stenson’s data demonstrates that, although all segments accept eclipsis fairly readily, there are three levels of resistance to lenition: those that do not significantly resist lenition (namely /p/, /b/, /m/, /k/, and /g/, which lenite in 75% or more of
tokens), those that resist it (/t/, /d/, and /s/, which lenite in 30-50% of tokens) and those that resist it strongly (this last group consisting only of /f/, which lenites in only 6% of tokens). I propose an Optimality Theory analysis to account for these differences, positing that the intermediate class of consonants resist lenition because they violate a high-ranked faithfulness constraint (namely IDENT-PLACE) in the process of mutating into their lenited forms. I further posit that /f/ resists lenition even more thoroughly because the mutation violates an even more highly-ranked constraint which prohibits consonant deletion. Also included throughout the paper is relevant discussion of phonological and sociolinguistic features of Modern Irish and how they intersect with the research question at hand. This paper is meant to be a sort of proof-of-concept that could guide further research, perhaps including novel analysis of conversational data to see if Stenson’s results are reproducible and if new results continue to conform to my conclusion here.