530
Berichte und Kommentare
Anthropos 85.1990
Austronesian Root Theory
Paul Geraghty
I was genuinely amazed by a discovery made while
working on this review-article of Robert A. Blust’s
“Austronesian Root Theory” (1988). Having de
cided to consult Otto Christian Dahl’s “Proto-Aus
tronesian” (1973), the text I was reared on as a
young postgraduate student at the University of
Hawaii, I scanned the bibliography to find not a
single reference to the works of Robert Blust. Aye,
indeed, that was another era. Such has been Blust’s
dominance of Austronesian comparative studies
over the past decade and a half that it is very rare to
find a scholar in the field who has not had cause to
refer to a Blustian reconstruction, or been goaded
into action by a piece of Blustian presumption
in his neck of the woods. It was Blust who first
proposed and then elaborated on the currently most
widely accepted subgrouping of the Austronesian
language family, and his passion for devouring da
ta from every comer of what is arguably the largest
and most diverse language family in the world has
yielded a body of reconstructions that surpasses in
every way the previous standard, published fifty
years ago by the father of Austronesian studies,
Otto Dempwolff. This fascinating book, which
is based on those reconstructions, demonstrates
for Proto-Austronesian (PAN) a level of linguistic
analysis between the phoneme and the morpheme,
and very throroughly explores the repercussions
for general linguistic theory.
Blust defines the Austronesian root in the
first chapter. It is neither the familiar “root” of
Indo-European, nor that of Semitic, but apparently
a peculiarly Austronesian phenomenon. It is “an
entire CVC (or, rarely, CV) syllable in which there
is no limitation on the filling of consonant and
vowel positions apart from the general morpheme
structure constraints of the language ... [it] may
occur as an independent form, but only if it is
onomatopoetic ... except when it occurs as an
independent form, [it] is always the last syllable of
a disyllable or longer word ... not all morphemes
contain a root ... some roots exhibit a pattern of
vowel variation which appears to be correlated
with semantic gradation .,. Most significantly,
analysis of the Austronesian root suggests that
there is a level of language structure intermediate
between the phoneme and the morpheme which
has been generally overlooked by linguists ... [it]
is an analytical product of recurrent association,
isolated through a lexically pervasive sound-mean
ing correlation in which neither the recurrent ele
ment nor the residue is independent, and in which
only the recurrent element has meaning. Opera
tionally, then, the Austronesian root bears a closer
resemblance to the phonestheme (English gl- in
glow, gleam, glimmer, glitter) than to the ‘root’
as previously discussed either in Semitic or in
Indo-European linguistics” (1-2). The theory is as
yet language-family-specific, but may eventually
be usefully applied across language families to a
class of sound-meaning associations which can be
identified in the absence of paradigmatic contrast.
Chapter 2, “The Morpheme: Two Operational
Definitions,” is a discussion of the definition of
the term “morpheme.” Bloomfield alone seems to
have defined it by recurrent association (rather than
paradigmatic contrast), and thereby admitted as
morphemes what later came to be termed “pho-
nesthemes” - isolable portions of morphemes that
have meaning but cannot occur independently, and
leave a residue that is itself meaningless. The
Austronesian root is a kind of phonestheme. Al
so included under this rubric are certain sub-root
phonesthemes of Austronesian (see below).
Chapter3, “A Synoptic History of Austrone
sian Root Theory,” leads the reader back to the
origin of Austronesian root theory in a 1916 paper
by Brandstetter, and details the characteristics of
the root, and means of forming word-bases from
it, as outlined in that paper.
Chapter 4 is a critique of Brandstetter’s theo
ry. While agreeing that the theory is “built around
a core of valid observations and explanations,”
Blust argues nonetheless that Brandstetter “at
tempted to extrapolate well beyond what the data
legitimately appear to support, and in so doing he
unwittingly discredited his theory in the eyes of
many later researchers” (10). Blust lists six major
weaknesses:
1. Many pseudo-roots were reconstructed as
a consequence of Brandstetter’s lax control of
chance, for example his requiring only two inde
pendent witnesses (Blust requires four), and not
taking into account phonological mergers.
2. Similarly, although he specifically cau
tioned against assuming that an attested monosyl
lable derives from an original monosyllable (since
an original consonant may have been lost, e.g.,
in Old Javanese), there are instances of his doing
precisely that.
3. Brandstetter repeatedly used “cognates” to
establish roots, when what is required is a recurrent
partial in “unrelated” forms.
4. Perhaps the fundamental cause of these
“errors of enthusiasm” was a belief - suggested