MVS -> OP stop series ?

[Koontz John E]

On Sat, 14 Feb 2004, Rory M Larson wrote:
> MVS had [p], [t] and [k], as well as [h] and glottal stop [?].
> [h] could combine with the three oral stops on either side;
> [?] could immediately follow them (but never precede them?).
> Thus, we have four series of stops:
>
>   [p]       [t]        [k]
>   [hp]      [ht]       [hk]
>   [ph]      [th]       [kh]
>   [p?]      [t?]       [k?]

Correct, though there are schools of linguistics that get really fussy
about thinking of hC, Ch and C? as combinations.  You might flunk out of
MIT for saying this wrong at the wrong moment.  Fortunately, in many cass
it appears that "combine" does explain the historical situation in Siouan.

> We also have double stop clusters, like [pt] and [kt], which
> have been reduced to single stops outside of Dakotan.  (Do all
> six possible combinations occur?)

Glossing approximate

pp OP ppase A1 (< base 'cut by pushing')
pt Da pte, OP tte 'bison (cow)'
pk Ma pke, OP kke 'turtle' (this one is a bit iffy, see next line)
pk OP kkaNbdha A1 (< gaNdha 'want'), but ppaghe A1 (< gaghe 'make')
tp Da nakpa, natpa, OP nitta, Wi naaNNc^awa 'external ear'
tk Da yatkaN, OP dhattaN 'drink'
kp Da kpaza, tpaza, OP ppaze, Wi (ho)kawas 'darkness'
kt Da ktA, OP ttE IRREALIS

> In Dhegihan, the stop retained is normally the second of the two; thus
> [pt] and [kt] both become [tt].  This stop is normally held long, or
> tense.

T wins if it's present, fore or aft, and this has been argued to be due to
metathesis.

> In Dakotan, the pre-aspirates [hp], [ht] and [hk] merge with the
> corresponding post-aspirates [ph], [th] and [kh].  (Right for which
> is pre- and which is post- ?)

Da shifts *th to h in initial position in verbs.

> In OP, the [hp], [ht] and [hk] drop the [h] and have the stop held long
> and tense: [pp], [tt] and [kk]. In addition, the plain stops [p], [t]
> and [k] are voiced to help distinguish them from the tense series,
> becoming [b], [d] and [g].

Correct.  In SC clusters C is essentially unvoiced.  In Cdh clusters it is
voiced.  S = s/s^/x; C = stop.

> In OP, the original glottal stop [?] is lost, and [k?] => [?], a
> new glottal stop set.  [t?] and [p?] are retained.

Many *ph become h.