Here is a link to “Making Sense of Shifty Changes: The Role of Phonetic Analogy in Vowel Shifts” by David Durian and Brian D. Joseph, the first of several presentations I am posting here from the “New Perspectives on Vowel Shifting” panels held at NWAV 40. Note–This version is an updated and somewhat more detailed version that what we actually presented in person at NWAV 40.
Here is our abstract:
The idea that diachronic development and cross-generational incrementation of vowel shifts
proceed via processes of rule extension/rule generalization has a long history (King, 1969;
Labov, 1972/1994/2010). Recently, linguists have questioned whether the generalization process
involved might better be classified as analogical, since parallel shifts usually develop and
progress in real world observation of actual speakers’ shifting vowel systems (Gordon, 2001;
Boberg, 2005; Durian, 2009).
Accordingly, we explore here whether this generalization process in parallel shifts is rule
extension/generalization or analogy, specifically phonetic analogy (Schuchardt, 1885;
Vennemann, 1972). Conceptually, phonetic analogy is a reasonable alternative to “extension”,
since “extension” is essentially an analogical likening of, say, a mid vowel to a high vowel.
As the empirical basis for our discussion, we present results of a quantitative analysis of two
parallel shifts (fronting of /uw-ow/; backing of /ae-E-I/) found among Columbus speakers. We
offer real and apparent time change trends from the vowel systems of 62 speakers, born 1896-
1990 and evenly representing social class (middle/working), sex, and generational cohort.
Our analysis shows speakers making use of phonetic analogy during parallel shift development/
This process involves three steps, during which linguistic relationships, such as contrasts between
combinations of place and voicing features which occur as contrastive sets, function as either “models” or “clones”: a) initial extension of a “model” set of relationships to another “clone” set; b) refinement of the modeled set relationship applied to the clone set; c) increasingly “generic” generalization of the refined clone set relationship affecting additional clone iterations.
Data from both parallel Columbus shifts exemplifies how each step occurs. Further, we explore application of this model to other vowel shifts, e.g. the Canadian and Northern Cities chain shifts. Ultimately, we conclude phonetic analogy is the best way to classify the overall generalization process in vowel shift development/incrementation, rather than rule extension/generalization.