Ma'a/Mbugu

[bwald]

----------------------------Original message----------------------------
We are now considering the issue of productive non-Bantu morphology in
Ma'a.  Sally notes:

>   Christopher Ehret identifies at least one causative suffix in
>Ma'a as of Cushitic origin; Mous cautions that Ehret's etymologies
>need careful checking, but in any case it doesn't appear to be a
>Bantu suffix, and Ehret says that it is -- or was, ca. 25 years
>ago -- very productive.

The only such causative suffix that I know for Mbugu is -ija, which
corresponds well to
Pare -ija (or -izha) and has a Bantu etymology.  That is not to say that it
could not also
correspond to a non-Bantu (probably Cushitic causative ?*-iyya). [Such
things happen,
cf. English-based creole "se", either English "say" or Akan.]  One earlier
observer,
E.C. Green, noted that when a simple verb ends in -a, e.g., hala (be
parched) the caus is
added as follows; hal-ija.  This is the Bantu way.  However, when the verb
ends in
another vowel, that vowel is preserved, e.g., hletu (lessen intr) cause
hletu-ja.  This is
not the Bantu way, because Bantu simple verbs always end in -a.  None of
these verb
roots are Bantu, but if hletu were (which it isn't) hletu-ja would
presuppose the Pare
simple verb *hletu(l)a (Pare lost etymological intervocalic -l-).

Next,
>   The pronominal possessive suffixes are certainly inflectional
>(at least that's how such affixes are generally analyzed in languages
>that have them), so indeed there is evidence of Cushitic inflection
>in (earlier) Ma'a.

In both Ma'a and the relevant Bantu languages (as general in North/Central
East Bantu)
the independent pronouns and possessive "suffixes" are not (all) identical
in form, e.g.,
Pare/Shambaa/Swahili/etc 1s mi- but poss -ngu.  Ma'a has (or had)
independent 1s ani
and possessive gho (or xo) [the reported phonology is velar but otherwise
varies a lot].

In relevant Bantu possessive constructions, the possessive suffixes are
affixed to the
morpheme -a- 'of', e.g., -a-ngu "my/mine".  In Ma'a they are not.   They
seem to be
treated as *independent* words.  There is no "linking" vowel -a-, as there
would be in
Bantu.  In addition, at least for non-human head nouns, there is
no concordial class marker, as there would be in the relevant Bantu
languages, e.g.,
Ma'a: ihle *gho* (name *my*), cntr. Pare:  jina *l-a-ngu* (name *it-of-my*),
where the l- is the class prefix for the class of jina 'name'.  The form of
the class prefix would
vary depending on the class of the head noun, but there would be one in any
case.  Ma'a
may be following a Cushitic construction (I don't know), but it is
certainly not following a
relevant Bantu construction.  The interesting parallelism is simply that
there are morphologically
distinct sets for possessive pronouns in both (relevant) Bantu and Ma'a.
Beyond that, the
morphology of the Ma'a possessive pronouns is word-like, not inflectional
in any decisive
sense.

Finally, about the possessives, in Ma'a the human ones are treated like
Bantu nouns
(or most descriptive adjectives).  They take the nominal concords, NOT the
ones used in
the relevant languages with possessives, e.g., Ma'a MU-gho (class.1-my) =
mine (e.g.,
the child is *mine*).  This is like Pare nominal (or adjectival) concord:
MW-eza (class.1-tall)
= tall (one) (e.g., a *tall* person), not like possessive concord W-a-ngu
(class.1-of-my) =
my/mine (e.g., the child is mine).  The difference is between the
concordial forms *MU and
*YU, both class 1 (typically human singular).  With possessives the
concordial form should
be *YU (in relevant languages) not *MU.

[whether they are used attributively or as predicates doesn't affect the
distinction here.]

>   Maarten Mous has suggested (p.c. 1993) the possibility that the
>pre-bantuized Ma'a language originated as a mixture itself, the
>language of a people coalesced from an escaped group of cattle
>herders who had been (semi-?)enslaved by the Masai.  This scenario
>could explain the lexical mixture in Ma'a.  But, as Mous notes, it's
>hard to imagine a way in which this hypothesis could be tested, given
>the total lack of social information about the Ma'a people from the
>period before they arrived in the Pares.

That seconds my point about the importance of social information.  We can
hope at
this point that languages evolve in a way that is so constrained (if we knew the
constraints) that eventually we might be able to extrapolate from similar
situations (if they
become recognised) for which we can get more evidence.  (In fact, if
linguistics lasts
long enough maybe a comparable situation will arise again somewhere, pace
the guys who
think everyone is going to end up talking "post-modern English".)

Lastly,
>P.S.  I don't think there's general agreement any more that Sandawe
>      and Hadza belong to the Khoisan family.  Bonny Sands' recent
>      UCLA dissertation failed to find solid support for the
>      hypothesis.  (For that matter, there seems to be considerable
>      doubt among Khoisan specialists that even the southern members
>      of the proposed family are related to each other.)

That is right.  Most experts in the field never accepted Greenberg's
classification of
Sandawe (or Hadza) is Khoisan, i.e., genetically related to the Southern
branches --
not to mention doubts about them being related to each other, e.g., Khoi
and San.
Khoisan is a label of convenience.  The typical issue remains whether they
are so
deeply related that it is hard (or impossible) to demonstrate, or whether
the typological
features they share, e.g., "clicks" and "gender" systems (hence Meinhof's
"Hamitic"),
may be an old speech area phenomenon since overlaid with Cushitic (in the
North) and
Bantu everywhere.  Sandawe, more than Ma'a, is in a unique African area
where members
of Greenberg's "four" major families are spoken.

P.S. There's more to be said about the details of how Ma'a adjusted to
Bantu grammar,
and where it stopped short (in the past), and I look forward to how that
will be treated in
the future.