10.1165, Qs: ..., Word Units

[Koontz John E]

> Date:  Thu, 5 Aug 1999 13:04:25 +1100
> From:  Alexandra Aikhenvald <Sasha.Aikhenvald at anu.edu.au>
> Subject:   Phonological and grammatical words
>
> We are studying languages for which it is appropriate to establish separate
> units of phonological word and grammatical word.  These sometimes coincide
> but do not always do so. Can you direct us to reliable descriptions of
> languages which recognise these two types of word, with explicit criteria?

I felt forced to make this distinction (approximately) in order to
simplify the description of Omaha-Ponca, a Dhegiha Siouan language.  I
actually hit on the concept though in dealing with Dakotan, so I think it
is a generally useful one in Mississippi Valley Siouan.  I'm not sure if
one can get any milage out of it in the other two or three branches of
Siouan proper.  Unfortunately, I cannot direct you to my dissertation as
it is a long way from done.

The general idea, however, is that verb words in particular often glue
together several smaller verb words.  These smaller words are always
either invariant particles or verb stems inflected by prefixation,
including in the latter class the patterns involving the so-called
locative prefixes, which embed within the pronominal string.  Typically
the inclusive pronominal precedes the locative, while the first and second
follow.  Third is zero.  There are some complications of this
characterization of the locatives in both Dakotan and OP, but it's close
enough.  Phonological words that are complex verbs will have several
inflected segments in a row, or a particle followed by an inflected
segment.  If one tries to treat these complex verbs as grammatical words
one has to write a much more complicated grammar than if one admits that
some phonological words are made up of several sequential grammatical
words.

Complex phonological words can arise lexically or syntactically.  As
examples of the former, the outer instrumental stems consist of a particle
followed by a grammatical verb, and some verb stems simply require two
particular simple stems in a row, e.g., Omaha-Ponca 'to want', which
involves two separately inflected substems:  A1 + gaN=dha => kkaN=bdha;
A2 + gaN=dha => s^kaN=na.  Note that inclusives only occur once in a
phonological word:  A12 + gaN=dha => aNgaN=dha.  An example of a
syntactically arising form would be a causative (the subordinated stem
becomes a particle), a negative (the negative enclitic has a defective
inflectional pattern), or an intensive or habitual (both inflected
enclitic stems).  Motion verbs engage in complex patterns of compounding
that might be seen as an inbetween case.

Interestingly, in some derivational patterns lexical complex phonological
words are treated as a single chunk, e.g., a reflexive of an outer
instrumental stem may treat the particle plus stem inside the reflexive as
a single grammatical piece.

While I am hesitant to express an opinion on the applicability of this
concept outside of Mississippi Valley Siouan, I know that Randy Graczyk,
in his dissertation on Crow found it convenient in some contexts to treat
whole phrases (of several phonological words) as incorporated into certain
verb stems, or, in phonological terms you could see the superordinate verb
as an enclitic of the last word of the phrase.  I guess you could say the
same thing of postpositions in Omaha, but without inflection of the
enclitic there's not much to be gained in traditional terms by calling the
phrase incorporated, unlike the Crow case.  A possible parallel in Dakotan
many conjunctions (postposed) and postpositions engage in a pattern of
final vowel truncation reminiscent of what happens with incorporated and
reduplicated "underlyingly C-final"  stems.