formally-determined morpheme order

[Jeff Good]

Dear Fritz,

I am not aware of a good case that closely matches your example language, that is where the constraints on ordering are stated in primarily phonological terms (though see one proposal below). However, Larry Hyman, in a number of papers, has argued that four of the Bantu valency-changing suffixes are subject to an ordering template by virtue of their morphological identity alone. A copy of the paper where this is argued for in most detail is here:
http://www.linguistics.ucla.edu/people/hayes/205/Readings/Hyman_SuffixOrderingInBantu_2002.pdf

If you accept his analysis, then I think it is the kind of case you are looking for.

You may also want to look at this dissertation by Mary Paster entitled, "Phonological Conditions on Affixation":
http://pages.pomona.edu/~mp034747/paster_dissertation.pdf

Her section 5.4, on possible cases of phonologically-conditioned affix order, would be most relevant. Of particular interest here is the Pulaar case, where it has been proposed that affix order is driven by a sonority condition describable as something like T > D > N > R, where the symbols correspond to suffixes containing those consonants (or something along those lines). There are alternative analyses (see Paster's dissertation), but this is probably the most striking proposal of formally-driven ordering for more than one or two affixes that I've seen. (Mary Paster's publication page also has relevant papers, http://pages.pomona.edu/~mp034747/papers.html.)

Best,
Jeff


On Feb 27, 2014, at 3:22 PM, Frederick J Newmeyer <fjn at u.washington.edu> wrote:

> Dear Funknetters,
> 
> I'm looking for an example where a language has three or more bound morphemes in a fixed order, but where the order among them cannot be characterized in semantic/pragmatic terms without loss of generalization. (I am including scope and notions such as 'relevance to the root' as part of 'semantic/pragmatic'.) Here is a hypothetical example of what I am looking for:
> 
> Some words in Language L consist of a root followed by three suffixes. The first suffix has to be bisyllabic, the second is pronounced [ka] and can be associated with a number of different meanings, and the third suffix has to be monosyllabic.
> 
> There is no language precisely like that, of course. But I am wondering if there are languages of roughly that sort, i.e., where the generalizations about morpheme position can be stated only in non-semantic, non-pragmatic, and non-discourse-related terms.
> 
> Thanks!
> 
> --fritz 
> 
> 
> Frederick J. Newmeyer
> Professor Emeritus, University of Washington
> Adjunct Professor, U of British Columbia and Simon Fraser U
> [for my postal address, please contact me by e-mail]
> 
> 
>