Areal Phonology

[Koontz John E]

Being laid up with tendonitis I've had an opportunity to read some things
I wouldn't ordinarily get to, and noticed a few points of interest in Paul
Proulx's article on Algonquian reduplication in IJAL 71.2.

1) Proto-Algonquian distinguishes two types of sandhi

   - [proto-Algonquian] inter-word insertion of *h.
   - within-word insertion of *y.

The first of these appears to be a discovery (or recent reinvestigation?)
of Dahlstrom (1997), exemplified in reduplication in Fox in

net-    es^a=        h-     es^awi
1s.subj reduplicator sandhi do.thus
'I do thus'

This pair reminds me of the Siouan tendency to

- CV1-V2CV => C-V2CV

on the one hand, where the boundary occurs in compounding, mostly, but
with morphologized expamples of

- insertion of *r intervocalically

in contexts such as combinations of locatives, the morphology of the
causative, and so on.

There are some signs in Mandan of V-?-V linkage, if not V-h-V linkage.
Carter has used this to explain some glottal stops in Mandan and has
suggested that it might also ultimately account for ejectives generally,
e.g., via a mechanism like CV1-?-V2 => C?-V2.  For example (mine of the
moment, not necessarily Carter's) te-e die + DECL might lead to t?e.
Similarly, it has been suggested somewhere - by Carter and/or Shaw, I
think - that declarative =? in some Dakotan dialects might have this sort
of source.

2) In connection with the former, Proulx notes that

"In Shawnee, Cheyenne, and Arapahoe, words which otherwise would be
vowel-initial are written with initial *h*.  However, many linguists
prefer not to write an initial aspiration as long as it remains automatic,
regarding it as sufficiently signalled by the space left between words ...
For example, Bloomfield (1962:3) states that in Menominee "initial vowels
often have an on-glide resembling *h*" which he does not write.
Similarly, in my description of Micmac inflection I wrote "there is
nondistinctive aspiration between (two) vowels, and before vowels in
utterance initial position."  I too did not write it initially, though I
did word-internally (Proulx 1978:5).

"These facts suggest the hypothesis that Proto-Algonquian may have had
predictable aspiration of a vowel-initial word.  In some daughter
languages aspiration renains predictable (e.g., Menominee).  In some it
may have been lost (e.g., Fox), and in some it changed to a glottal stop
(e.g., Potawatomi).  Reduplication with external sandhi inserts this
word-initial *h* at the [proto-Algonquian] word boundary between a
reduplicator prefix and a vowel-initial stem."

A footnote disposes of an anonymous reviewer's comment that word-initial
*h* is not written in pre-19th century Shawnee vocabularies though
English-speaking recorders might have been expected to to have written it.
Not, I suppose, any French-speaking ones?

While h-linkage is not attested in Siouan that I am aware of, I thought
that the presence of automatic initial aspiration in Menominee was
interesting in light of the anomalous initial aspiration of verbs
(locatives and first persons) in Winnebago.  This aspiration, and the
presence of initial h with Ioway-Otoe first persons, has always puzzled
me, in the sense that I couldn't entirely predict when it occurred.

For example, Wi ha-, IO ha-, OP a-, Da wa- for first person
agent is not a regular correspondence, though the initial do hold across
essentially all first person pronominals, e.g., Wi hiN-, IO hiN-, OP aN-
(dative iN-), Da maN- (reflexive possessive miN-) for the patient forms.

Winnebago has nouns without h-, e.g., aa 'arm', aap 'leaf', ii 'mouth', ii
'lid', and demonstratives ee 'the aforesaid'.  Also adverbs, e.g., aaki'
'on or at both ends or sides', and verbs, aaghi' 'be ready', aaz^(=)re 'be
open'.  The verbs considered to be *?-initial lack h, e.g., iNiN 'wear
over the shoulders' and uNuN 'do; make; wear'.  In fact, mainly it's the
three locatives and the first persons that get the aspirate.  The only
aspirated forms in IO are the first persons.

Given the general lack of *w => h shifts, and the absence of h- with
Dhegiha first persons, my suspicion would be that h- with first presons in
Winnebago was aspiration of (short?) vowel initial words, but that makes
the IO aspiration of first persons only somewhat anomalous.

The only other h'aspiration anomaly in Siouan that I know of is the
behavior of 'day', e.g., Da aNp-e-tu, OP aNbe, Os haNpe, Wi haNaNp a
correspondence also seen with the indefinite pronominal (Da examples
moot), OP a-naN 'how many, some number', Os ha-naN, Wi hac^aN' 'where?'.

John E. Koontz
http://spot.colorado.edu/~koontz