Dixon's "The rise and fall of languages"

[Harold Koch]

----------------------------Original message----------------------------
On Australian languages:
A. Manaster Ramer wrote 9.5.98:
<...Australian-Pamanyungan (whose unity was proven years ago by Ken Hale)..
 
My understanding is that the Pama-Nyungan family was established in the
early 1960s on the basis on lexicostatistical percentages, backed up with
the implicit structural knowledge of the languages gained by the
classifiers O'Grady, Hale, and Wurm.>
 
Miguel Carrasquer Vidal writes, the same day:
<Dixon makes four basic assumptions:.....
3) Core vocabulary is not (universally) more resistent to change than
   "non-core" vocabulary.......
 assumption (3) is surprising:
        This [greater stability of core vocabulary] does appear to
        hold for the languages of Europe [..] and of many other parts
        of the world.  But it does not apply everywhere [..]  In
        Australia, for instance, similar percentages of shared
        vocabulary are obtained by comparing 100 or 200 or 400 or
        2,000 lexemes, from adjacent languages.>
 
In my opinion it has been demonstrated sufficiently by Paul Black that
assumption 3 is indeed valid for Australian languages (Black, Paul, 1997,
"Lexicostatistics and Australian languages: problems and prospects". In
Tryon, D. and Walsh, M. (eds), The boundary rider: Studies in Honour of
Geoffrey O'Grady. (Pacific Linguistics C-136) Canberra: Research School of
Pacific and Asian Studies, The Australian National University, 51-69).
 
Miguel Carrasquer Vidal further writes:
 
< One supposed (sub-)family, Pama-Nyungan, covers about 85% of the continent.
However, neither for Pama-Nyungan (apparently [p. 91] a
lexicostatistic construct), nor for Australian as a whole have family
trees or proto-languages been succesfully set up.  "It is possible to
establish low-level subgroups in Australia -- groups of from two to a
dozen or so languages that appear to have a close genetic
relationship", but the usefulness of the family tree concept as far
as Australia goes is apparently so low, that Dixon does not even
bother to mention the number of distinct genetic groups.>
 
Readers interested in further information on the historical-comparative
situation of Australian languages, including alternative interpretations to
those expressed in Dixon's Rise and Fall, might consult two books on
Australian comparative linguistics which  appeared last year:
McConvell, Patrick & Nicholas Evans (eds). 1997. Understanding ancient
Australia: perspectives from archaeology and linguistics. Melbourne: Oxford
University Press.
Tryon, Darrell and Michael Walsh (eds). 1997. The boundary rider: Studies
in Honour of Geoffrey O'Grady. (Pacific Linguistics C-136) Canberra:
Research School of Pacific and Asian Studies, The Australian National
University.
 
I excerpt below some of the views I expressed in each:
 
On Proto-Pama-Nyungan:
"It can probably not be said yet that pPN has been reconstructed in as much
detail as O'Grady (above) suggested is necessary before Arandic cognates
can be exploited. The number of reliably reconstructed words is still
pretty meagre. Moreover, we lack a handy list of reconstructions: there is
no etymological dictionary available (although O'Grady's comparative files
surely contain the makings of such a work). We still lack a proper subgroup
structure for the Pama-Nyungan family. While this situation prevails, we
have no certainty as to what spread of languages permits a set of cognates
to be reconstructed for pPN, as opposed to a subgroup of Pama-Nyungan...."
 
(Koch, Harold. 1997b: "Pama-Nyungan reflexes in Arandic languages". In
Tryon, D. and Walsh, M. (eds), The boundary rider: Studies in Honour of
Geoffrey O'Grady. (Pacific Linguistics C-136) Canberra: Research School of
Pacific and Asian Studies, The Australian National University, 271-302:
quotation pp 273ff)
 
 
A summary of the results of comparative methods applied to Aust lgs (the
rest of the article is an exposition of the methods of comp ling for the
benefit or archaeologists et al):
 
"Results for Australian linguistic prehistory
 
Little comparative linguistic research in the sense we have been discussing
it took place on Australian languages before the twentieth century. Father
Wilhelm Schmidt, working from Vienna at the beginning of this century,
managed to recognise some of the low-level subgroups (Schmidt 1919). But
these are always fairly obvious because of the masses of shared vocabulary
and grammatical forms.
 
        Capell (1956) compiled a set of words which are widely distributed
over the continent, and called this vocabulary "Common Australian", without
making overt claims about what level of proto-language it could be ascribed
to. Capell (1956, 1962) also distinguished typological groupings of
Australian languages, and presented scenarios whereby languages of one type
might be transformed into another type (see also Wurm 1972). It is to him
that we owe the classification of languages into prefixing and
non-prefixing types, and various kinds of noun-classifying types.
 
        A systematic survey of Australian languages in the late 1950s by
O'Grady, Hale, and Wurm resulted in a tentative genetic classification,
based largely on lexicostatistical data, of all the languages of the
mainland (O'Grady, Voegelin and Voegelin 1966, Wurm 1972, Walsh 1981). For
the application of the methods, see O'Grady (1960), O'Grady and Klokeid
(1969), Hale (1962). The terminology of this classification has been widely
used ever since, even though the classification of particular languages has
been changed and some linguists have expressed scepticism about the methods
used in arriving at the classification. One of the main results was the
finding that a large family, labeled Pama-Nyungan, extended over most of
the mainland excluding most of the Top End and the Kimberleys. Within the
northern area, a great degree of diversity exists; the classification
enumerated as many as twenty-eight families. A discontinuity is found in
Arnhem Land, where the Yolngu languages of Northeast Arnhem Land belong to
the great southern Pama-Nyungan family, unlike all their neighbours in the
north. Another significant finding was the relative uniformity of the
languages of Western Australia south of the Kimberley area. They were all
assigned to the Nyungic subgroup. Another widespread subgroup Pama-Maric
extends from the tip of Cape York Peninsula to as far south as the New
South Wales border.
 
        There has been considerable disagreement over the extent to which
shared vocabulary can be used as a guide to genetic subgroups. Those
responsible for the lexicostatistical classification regard the sharing of
a relatively great amount of basic vocabulary as at least a tentative
support for the positing of close genetic relations, in particular of
Pama-Nyungan (Wurm 1972, O'Grady  1979, Hale 1982). On the other hand,
Dixon (1970a, 1972a, 1980) and Heath (1981) have claimed that borrowing
between neighbouring languages can render the method of lexicostatistics
totally unreliable. As mentioned above, however, Black (1979) has proposed
a method to control for the lexical distortion caused by borrowing. Dixon
(1972a, 1980:255) has proposed that through borrowing adjacent languages
will over time reach a point of equilibrium, sharing 40-60% of their
vocabulary with each other, regardless of the closeness of their genetic
relation. Alpher and Nash (1984ms) argue that Dixon has overestimated the
proportion of vocabulary replacement that is attributable to borrowing, and
hence that the equilibrium level may be much lower than that claimed by
Dixon. In fact, cognate percentages of 10% or less are often found in Top
End languages. Dixon (1972a) also claims that the practice of tabooing the
names of the recently deceased (and similar sounding words) leads to a
relatively rapid replacement of vocabulary through borrowing. Arguments
presented by Heath (1979:409), by Black (1980) and by Alpher and Nash
(1984ms) suggest that the consequences of taboo for the recognition of
genetic relations are not as severe as suggested by Dixon. In spite of
these disputes, however, most linguists would agree that verbs are replaced
by borrowing much less easily than nouns and adjectives, and that a high
level of sharing of verbs is therefore a more reliable indicator of a close
genetic relation between languages than figures that group all kinds of
vocabulary. Personal pronouns and basic body part terms are also relatively
stable historically, whereas terms for flora, fauna, material culture, and
human classification are much more prone to borrowing (Breen 1990).
Furthermore, all linguists put more reliance on agreements in grammatical
forms than on the sharing of vocabulary items.
 
        The comparative method has been applied with success to reconstruct
the history of phonologically aberrant languages of Western Australia
(O'Grady 1966, Austin 1981), Cape York Peninsula (Hale 1964, Sutton 1976,
Black 1980, Dixon 1991), New England (Crowley 1976 and This volume), and
Central Australia (Koch To appear). In many of these cases, certain
languages had undergone radical sound changes while close relatives
remained unchanged. Thus the proto-forms for a subgroup often turn out to
be identical to forms surviving intact in other languages of the subgroup.
 
        We have a fairly good idea about Proto-Pama-Nyungan structural
features. Dixon (1970, 1980) has written on the phonology of the
proto-language-- although Crowley (This volume) reminds us that this
phonology has not been reconstructed by the strict application of the
comparative method. Evans (1988) has contributed further insights through
the application of the comparative method. Dixon (1980) and Blake (1979,
1988, 1990b) have reconstructed the main features of pronouns and noun
inflections; Alpher (1990) has done the same for verb inflection. O'Grady
and his students are working on Proto-Pama-Nyungan vocabulary
reconstruction (see O'Grady and Tryon 1990; Fitzgerald and O'Grady, This
volume); we are nevertheless far from being able to supply lists of
thousands of reconstructed Proto-Pama-Nyungan lexical items, comparable to,
say, Proto-Oceanic.
 
        It must be admitted that as long as so much remains unreconstructed
in lower-level subgroups, and the relation of these subgroups to one
another remains unclear, our picture  of Pama-Nyungan as a whole is still
rather incomplete. There is some doubt as to how, and whether, all of the
languages of southeast Australia fit into Pama-Nyungan (Evans 1988, Blake
1991:50-52). One researcher, Dixon (1980 passim), even remains unconvinced
of the validity of Pama-Nyungan as a genetic construct.
 
        A considerable amount of comparative study has been undertaken
recently on the northern language groups. Blake (1990b) has compared
personal pronouns across all the non-Pama-Nyungan languages, suggesting
that they give evidence that all these languages are related to one another
more closely than to the Pama-Nyungan languages. The postulated genetic
diversity in the north is being reduced: in place of the twenty-eight
families of the original lexicostatistical classification we may see
ultimately as few as eight or ten major genetic groupings among the
non-Pama-Nyungan languages of north Australia. The forthcoming works by
Dixon et al., Evans, and I. Green will make available much more of the
results of recent comparative study of these languages.
 
        No one can say much at this stage about Proto-Australian, the
assumed ancestor of all the continental Australian languages. It is now
widely accepted that Dixon's "Proto-Australian" reconstructions apply more
appropriately to Proto-Pama-Nyungan. Nevertheless Blake (1988, 1990b)
tentatively reconstructs a Proto-non-Pama-Nyungan set of personal pronouns,
which he finds to be relatable as "sisters" to the reconstructed
Proto-Pama-Nyungan set. This implies that both of these groups have a
common ancestor, which could be called Proto-Australian. Many of the verb
roots that Dixon (1980) discusses, but not his reconstructed inflectional
systems, may likewise go back to a Proto-Australian (Heath 1990; Alpher
1990; Alpher, Evans and Harvey To appear).
 
        It appears that all the languages of mainland Australia are
genetically related. The languages of Tasmania are doubtful, and there is
too little evidence for us ever to be sure whether they were genetically
related to the languages spoken on the mainland at the time of European
colonisation (for a discussion of the evidence, see Crowley and Dixon
1981). Whether there is a genetic relationship with languages of Papua New
Guinea and/or Irian Jaya remains to be demonstrated, although Foley (1986)
attempted a comparison. It should be mentioned that many North Australian
languages share typological features with some of the these so-called
"Papuan" languages (see Nichols This volume).
 
        Several interesting geographical discontinuities between
genetically related languages cry out for a historical explanation. As
mentioned earlier, the Yolngu languages of Northeast Arnhem Land are
separated from the rest of their Pama-Nyungan congeners. Yanyuwa, on the
Gulf of Carpentaria, is now accepted as belonging to the Warluwarric
subgroup of Pama-Nyungan (cf. section 5.2), whose other members are found
in West Queensland and the Northern Territory border area (Blake 1990a,
1990b). The Tangkic subgroup of languages at the head of the Gulf and on
offshore islands are separated from the rest of non-Pama-Nyungan. Finally,
the Barkly languages (Jingulu, Wambaya, Ngarnji, Kutanji) are separated, by
Pama-Nyungan languages and Wardaman, from their Djamindjungan relatives
near the Joseph Bonaparte Gulf coast (Chadwick 1979, 1984).
 
        Loanword studies have not been pursued very much as yet (but see
studies by Evans, McConvell, and Nash in this volume). Nevertheless Heath
(1978, 1979, 1981) has done an exemplary study of intense contact between
the Yolngu language Ritharrngu and its non-Pama-Nyungan neighbours Ngandi
and Nunggubuyu, and given many indications on how to do what he calls
"diffusional linguistics". An important study by Walker and Zorc (1981) of
loanwords from the Indonesian so-called "Macassan" languages has documented
the cultural influence of these foreign traders in Northeast Arnhem Land.
This study has been pursued further along the northern coast by Evans
(1992a). McConvell (1985) in a brilliant study was able to demonstrate the
likely locus and mechanism of the creation and diffusion of subsection
terms. Hercus (1972, 1987) has done pioneering studies of areal linguistics
in South Australia and southwest Queensland. In addition, Blake (1979:324)
and Dixon (1980) contain maps showing the geographical distribution of
various structural features, some of which may be due to areal diffusion
rather than genetic groupings. Dixon et al. (To appear) will include much
further discussion of diffusion among the Australian languages.
 
        Little if any work has been done on linguistic evidence for
proto-culture or the spread of cultural innovations. We await studies on
terminologies for such items as the woomera, returning boomerang,
didgeridoo, grinding and leaching techniques, the dog, etc. Almost no terms
of proto-culture or environment have been reconstructed for a deep genetic
level, except *kuya 'fish' (Dixon, p.c.). We could probably add *kana
'yamstick' on the basis of its wide distribution. Evans and Jones (This
volume) make some further promising suggestions for reconstructed tool
terminology. In addition, some bird names are very widespread; e.g.
 
 *kuruku, kuluku 'dove'
 *waaka  'crow'
 *tyitityiti     'willy wagtail'
 *kurrkurr       'boobook'
 
It is unlikely, however, that any inferences about early homelands or
population movements could be made on the basis of reconstructed bird
names. O'Grady (1990:86) reconstructs a proto-Pama-Nyungan *mungka
'anthill, termite mound', which also does little to localise the speakers
of this early language, since anthills are practically ubiquitous in
Australia."
 
(Koch, Harold. 1997a. "Comparative linguistics and Australian prehistory".
In P. McConvell & N. Evans (eds), Understanding ancient Australia:
perspectives from archaeology and linguistics. Melbourne: Oxford University
Press. 27-43; quotation 40-43)
 
Harold Koch, Senior Lecturer
Department of Linguistics
Faculty of Arts
The Australian National University
Canberra ACT 0200 Australia
 
Telephone:  (02) 6249 3203 (direct) / ..3026 (messages)
(overseas) 61 2 6249 3203
Fax:  (02) 6 279 8214
(overseas) 61 2 6279 8214
email:  Harold.Koch at anu.edu.au