Date: Saturday April 6, 2013
Contact: Tobias Scheer
Invited speaker: Patrick Honeybone, University of Edinburgh
Deadline for submission: November 15, 2012
Notification of acceptance: January 20, 2013
Submission of abstracts: EasyChair
In 2013, linguists celebrate the 100th anniversary of Ferdinand de Saussure's death. It is therefore an important year regarding the split of diachronic vs. synchronic study of sound patterns. Saussure is known for having introduced the synchronic perspective into the study of language with his Cours de Linguistique Générale, after having made ground-breaking contributions to diachronic phonology (e.g. in his Mémoire).
Some 60 years after the publication of the Cours, the field swung into the exact opposite extreme of the spectrum, compared with where Saussure started out from: in early generative phonology, all patterns had a synchronic explanation, albeit one which often mimicked the history in derivational terms.
More recently, some voices have proposed the virtual opposite of this, viz. that all explanation of patterns is diachronic. But even among those who agree that synchronic and diachronic explanations are necessary, there is no agreement where to draw the line, and no criterion could thus far be identified that would allow the linguist to tell, for a given pattern, whether it is the result of 1) synchronic phonological computation, of 2) synchronic non-phonological computation (allomorphy), or whether it represents 3) distinct lexical recordings.
The typical analysis will assume that regular and productive patterns are due to 1), whereas exceptions and non-productive alternations are the result of idiosyncratic history that today appears as 2) and 3). By contrast, usage-based accounts assume that the synchronic system is nothing but a list of exceptions: all regularities arise in diachronic development.
Another, related, aspect is the way in which diachronic evolution could, or should be 1) used and 2) represented in the synchronic computational system of phonology. It is obvious that there is no such thing as diachronic computation: no brain-based system takes an input of, say, the 14th century and computes an output of the 21st century. Computation is only synchronic.
So the question arises how innovation comes into being and, once it has occurred, enters the synchronic computational system: two widely held (and conflicting) views are based on acquisition (misperception) on the one hand, and on social group identity (sub-groups want to be different) on the other. Both are based on non- (or pre-) grammatical phonetic variation.
But even if computation is only synchronic, there are ways to implement diachronic processes directly in the synchronic system (and hence not relegating them to allomorphy or the lexicon). In The Sound Pattern of English, for example, the electri[k] - electri[s]‑ity alternation is made of two rules: t → t͡s /__i that is present since the 11th century, and one that takes t͡s to s without context, added later on in the development of English. In purely surface-based theories this is more difficult to do, but in the work of many, a theory like OT now is also equipped with a derivational component (including intrinsic or extrinsic ordering).
Finally, an issue regarding the usage and representation of diachronic events in phonological study are eventual unattested intermediate stages: through how many intermediate stages has an attested form gone that is related to an older attested form?
This diachronic distance is a relevant question for example when forms of the same etymological item that occur in different dialects are compared: an implicit assumption often is not only that there is a common ancestor, but also that the differences observed represent a single phonological process. This caveat is still more acute since there is no agreement as to what counts as a minimal (or atomic) diachronic change (called the quantum by Lass).
It may also be asked, in this context, what status diachronically related forms have that appear in typological surveys that are designed to show what phonological computation can and cannot do. For example in Greek, reconstructed *odwos turned into a later o:dos and is often used to demonstrate that this kind of compensatory lengthening, where the trigger and the target are separated by a segment, is possible. This is based on the assumption that speakers' knowledge was involved in this phenomenon, something that may need to be shown independently.
Presentations addressing the abovementioned issues, or related topics, are welcome at the workshop. It is assumed that they are informed of earlier debates regarding the diachronic question, namely in the context of the 70s, where the most serious challenger of the mainstream was Natural (Generative) Phonology.