Much of the typology for inserted vowels-i.e. those that are not present etymologically, but which surface in the acoustics-has come from cross-linguistic patterns identified perceptually by linguists. Crucially, though, inserted vowels may superficially sound the same, despite emerging at different stages of derivation. Inserted vowels may be present in the underlying representation as vowels that have become fossilized over time (i.e. lexicalized vowels), in the surface representation by way of a derived process (i.e. epenthetic vowels), or in the phonetic implementation as a result of gestural alignment (i.e. excrescent vowels). In a modular feedforward framework, in which phonology feeds phonetics, phonological processes can display gradience typically associated with phonetic processes. Determining what level of representation an inserted vowel originates from, then, can be challenging and may result in misclassification. Misclassifying an inserted vowel can have consequences for phonological theory, since the theory is modeled to generate the typological patterns found in natural language production. A vowel that has been misclassified may appear to be a typological exception, and this exceptional behavior can result in proposals for new categories of inserted vowels which do not fit the phonology-phonetics dichotomy, or revisions to the modular feedforward framework to incorporate some degree of overlap between the phonetics and phonology modules. It is important, then, that the methodology used to classify inserted vowels as originating in the surface representation, for example, do so by excluding the possibility that the vowel did not originate in the underlying representation or from gestural alignment. In this dissertation, I utilize an under-documented case of vowel insertion in Scottish English as a case study to distinguish between lexicalized, epenthetic, and excrescent vowels in the acoustic signal. Using data collected in two language production experiments, I assess the inserted vowel's phonological and phonetic properties against two sets of diagnostic criteria-one which I use to establish that the vowel is phonological, and another which I use to establish that the vowel is epenthetic. These diagnostic criteria are applied via a process of elimination, in which I exclude the possibility that the vowel is excrescent to establish that it is phonological using the first set of diagnostic criteria, and subsequently exclude the possibility that the phonological vowel is lexicalized to establish that it is epenthetic using the second set of diagnostic criteria. The language-specific findings, novel diagnostics and exclusion process, and patterns for epenthetic vowels presented in this dissertation serve to supplement Scottish English phonology, improve the methodology available to phonologists investigating the origins of inserted vowels, and contribute to the typology of inserted vowels in a modular feedforward framework.
Read
