Note: The following article is a contribution to the "discussion" which has raged the past six years on the origin of the Finns, the ontology of the Uralic proto-language and the methodology of Uralic historical linguistics - i.e. the "family tree" discussion. I have written it partially because in my review of Künnap's "Breakthrough in present-day Uralistics" (Tartu 1998) in Linguistica Uralica 1999:4 I did not go into the matter as deeply as I would have liked - it being a book review, after all - and because my own ideas on the subject have developed a little. There are a few reasons why I publish it on the Internet rather than on paper. First of all, I dislike the idea of trees getting killed to print a contribution to a discussion which I believe is unproductive at best. Second, I have very little substantial to add to excellent critical articles of Esa Itkonen, Johanna Laakso and Eberhardt Winkler. Not printing it has also enabled me to go on a bit of "quoting-madness" on my own - this is a word of warning. Those of you who are still interested can read on.
16 Jan. 2002: I added provisional english translations of quotes in Finnish and Estonian. MdS
[1]
URALISTS AGAINST HISTORY
Merlijn de Smit
april 9, 2001
isolintu@hotmail.com
Mundus dicitur quasi undique motus
est enim in perpetuo motu
Honorius of Autun, Imago Mundi 1:1
BACKGROUND
In my opinion, the body of ideas proposed in recent years by a number of researchers in different scientific disciplines, most
notably by Kalevi Wiik and Ago Künnap, which have lead to the publication of a number of books and to an occasionally
sharply polemical discussion in different journals of Uralic linguistics, concern two different hypotheses, which are not
interdependent - in fact, I believe it would be imprudent to regard these two are two sides of the same "innovationist" coin.
The first of these hypotheses is a new hypothesis on the location of the Uralic proto-language in space and time and on the
processes by which it has dispersed. The second is a new view on the nature of the proto-language and the way in which
language is transferred through time itself, and acceptance of the latter would bear momentous consequences for the
methodology of historical linguistics.
As Esa Itkonen (1999: 88) remarked, it is inaccurate to speak of two camps in present-day Uralistics since the "traditional"
camp is not a monolith. It is almost impossible to define "traditionalist" Uralistics except perhaps by its acceptance of the
comparative method. As for the question of the location of Proto-Uralic in time and space, traditional comparative
linguistics has generally operated with a "bottom line" of some 6,000 years before Christ, since it is generally considered
impossible to discern older genetic relationships between languages. Locating the place of origin of a proto-language is
riddled with difficulties and pitfalls. Its different purely linguistic methods, like linguistic paleontology and the identification
of ancient layers of loanwords are often inconclusive in themselves. Linguistic paleontology bases itself on the semantic
reconstruction of lexemes, the interpretation of which (is a 'horse' a wild horse or a domesticated horse? Is it your own horse
or the horse of some neighbour population?) is often problematic (Renfrew 1987: 80-86, Beekes 1995: 47-48,
Sims-Williams 1998: 509) - which doesn't mean, of course, that hypotheses arrived at by linguistic paleontology should be
dismissed out of hand . The identification of ancient layers of loanwords, for example PIE loanwords in Proto-Uralic and
Proto-Finno-Ugric, and drawing conclusions from it regarding the location of the proto-language, essentially makes one
hypothesis on the location of a proto-language dependent upon another (Sims-Williams 1998: 510).
[2]
A massive general problem in sifting the linguistic evidence for a location of a proto-language is that the evidence at hand is
probably incomplete - it being based only on the languages which have survived (1).
The location of a proto-language, and its dispersal, in time is often based on the identification between archaeological cultures and languages - based on the principle that archaeological cultural continuity presupposes linguistic continuity (Makkay 1992: 196), and the assumption that, in the words of Christian Carpelan (2000: 8) "[...] there is a strong possibility that an archaeological culture, defined by ceramics, represents also a common vernacular.". As Robert S.P. Beekes (1995: 45) puts it: "Linguistic information offers us no basis for determining the moments of time at which the Indo-European peoples began to inhabit the areas which would later become the areas where they settled. Evidence for this must come from archaeology [...]".
Thus the arrival of Uralic languages in Finland is often equated with the arrival of comb ceramics around 4,000 years B.C.
(Carpelan 1996: 14, Kallio 1997: 123), and early Indo-Europeans are seen as the bearers of the Sredny Stog culture in
Southern Russia around 4,500-3,500 B.C. (Beekes 1995: 50). This approach is problematic. The assumption of a
one-to-one correspondence between an archaeological culture and a reconstructed language - and that the spread of an
archaeological culture implies the spread of a language - is dangerous (Renfrew 1987: 86-94, Sims-Williams 1998: 511),
since it contains assumptions on the nature of an archaological culture (linguistically homogenous) and its spread
(co-occurring with linguistic expansion and possibly expansion of a population) which are not necessarily based on the
evidence provided by archaeology. The danger of drawing inferences about the one based on evidence provided by the other
was aptly commented by Renfrew (1987: 19): "It is perhaps reasonable that the historical linguistics should be based upon
the archaeology, but that the archaeological explanation should simultaneously be based upon the linguistic analysis gives
serious cause for concern. Each discipline assumes that the other can offer conclusions based upon sound independent
evidence, but in reality one begins where the other ends. They are both relying on each other to prop up their mutual thesis."
The new approach in putting together archaeological, linguistic and now also genetic evidence to attain a big picture of the
populations and languages of prehistoric Europe, hailed by the "innovationist" Uralists (Künnap 1998a: 40-49) concerns
time-depths generally considered out of reach for historical linguistics. Whereas Milton Nuñez dates the colonization of
Finland by Uralic speakers to the end of the last ice age, around 10,000 B.C. (Nuñez 1997: 56), hypotheses of the linguistic
situation in Europe have been made using even bigger time-depths, up until 40,000 B.C. (Künnap 1998a: 36). The argument
that it is impossible to make any statement of the linguistic situation using these time-depths, and that any hypothesis would
almost rely entirely on archaeology, has been made (Larsson 1995: 34, also Winkler 1999: 248).
[3]
However, it has also been pointed out, for example by Colin Renfrew (2000: 14-15) that the linguistic "borderline" of
6,000-8,000 years after which genetic linguistic relationships can no longer be discerned is quite arbitrary and not, in fact,
based on any linguistic evidence. However, rejecting glottochronology and principally accepting a substantially older timing
of the proto-language does not mean that no limits exist. Although the timing of proto-Uralic at about 10,000 B.C. may be
acceptable, this does not mean automatically that making statements about "pre-Lapps" around 40,000 B.C., thus long
before any imaginable timing of PU, is acceptable.
In a recent article, Colin Renfrew summarized the aim of the new approach as follows: "The wider aim, however, must be to
establish a number of general principles relating language change to demographic, social and economic change." (Renfrew
2000: 11), "If, however, we take what I would regard as the optimistic view, and accept the proposals made earlier for
greater time depths, we then reach the possibility for some correlations or equivalencies between linguistic events and
climatic sequences and socio-economic processes" (Renfrew 2000: 24). Thus the expansion of languages are no longer
equated with the expansion of archaeological cultures, but rather seen against the background of some of the sweeping
demographical and economical changes of prehistoric Europe, the most important being the Neolithic revolution. As for the
expansion of the Uralic languages, the recession of the Ice cap covering Scandinavia and Finland and the subsequent
movement of big game into the newly uncovered land would be the relevant factors facilitating such an expansion (Julku
2000: 126-127).
Nuñez' model for the expansion of the Uralic languages into Finland, be it right or wrong, does not implicate any
consequences for the nature of the Uralic proto-language, for the methodology of historical linguistics or indeed for the
status of the language family tree - in fact, Pekka Sammallahti, one of the first Uralists to accept Nuñez' model in his
paperLanguage and Roots at the Jyväskylä congress in 1995, did so while defending the "traditional" view on the existence
of a Uralic proto-language and its subsequent break-up into daughter languages (Sammallahti 1995: 143-144). It is
principally possible to accept the "innovationist" model about the location of Proto-Uralic in time and space while rejecting
the "innovationist" conception about what is essentially the way language moves through time. Whereas I believe the new
conception about the time and place of Proto-Uralic is interesting, it is beyond my expertise to either accept or reject it. This
is not so of the new conception about Proto-Uralic and the Uralic family tree. In my opinion, the new conception is
destructive of the methodology of historical linguistics, entails a static and anti-historical view on language and, ultimately,
rather than a breakthrough in Uralistics, resembles more a recession of Uralistics to a pre-scientific age. I will elaborate on
my objections below.
[4]
UNDIQUE MOTUS.
In his recent article on tree models, Urmas Sutrop (2000: 165) writes: "The belief that both nature and human society
develop according to certain laws was characteristic of the intellectual climate of the 19th century. August Schleicher [...]
wrote, for example, "What Lyell carried out for the natural history of Earth the same did Darwin for the natural history of
life on earth" [...]. Social scientists thought that even human history follows some law of evolution." What Sutrop refers to
is, I believe, the defining breakthrough in 19th century science, to which many current scientific disciplines, among them
historical linguistics, owe their existence: the breakthrough of the conception that the material world, and thus the object of
scientific research, is not a creation of God, static or in decline from an ideal, pristine state, but subject to constant change -
and that this change happens according to laws which may be uncovered by the human mind.
As Friedrich Engels wrote in Ludwig Feuerbach und der Ausgang der klassischen deutschen Philosophie, published in the
middle of this scientific revolution: "Die alte Metaphysik, die die Dinge als fertige hinnahm, entstand aus einer
Naturwissenschaft, die die toten und lebendigen Dinge als fertige untersuchte. Als aber diese Untersuchung so weit gediehen
war, daß der entscheidende Fortschritt möglich würde, der Übergang zur systematischen Untersuchung der mit diesen
Dingen in der Natur selbst vorgehenden Veränderungen, da schlug auf philosophischem Gebiet die Sterbestunde der alten
Metaphysik. Und in der Tat, wenn die Naturwissenschaft bis Ende des letzten Jahrhunderts vorwiegend
sammelndeWissenschaft, Wissenschaft von fertigen Dingen war, sie ist sie im unserm Jahrhundert wesentlich ordnende
Wissenschaft, Wissenschaft von den Vorgängen, vom Ursprung und der Entwicklung dieser Dinge und vom
Zusammenhang, der diese Naturvorgänge zu einem großen Ganzen verknüpft. Die Physiologie, die die Vorgänge im
pflanzlichen und tierischen Organismus untersucht, die Embryologie, die die Entwicklung des einzelnen Organismus von
Keim bis zur Reife behandelt, die Geologie, die die allmähliche Bildung der Erdoberfläche verfolgt, die alle sind Kinder
unseres Jahrhunderts." (Engels 1888: 52). This shift of paradigm had great consequences for biology (Darwin), social theory
(Marx) and, of course, historical linguistics. Thus biology moved forward from merely classifying species (Linnaeus)
towards researching their historical development (Darwin), instead of regarding them as unchangeable creations of God.
Likewise, linguistics progressed from descriptive grammar towards researching language from a historical perspective - no
longer regarding languages merely as the offspring of Babel or the degenerated exponents of the great classical languages,
but as objects of birth, constant development and extinction. This dichotomy between a view regarding the world as static
and unchanging and a view regarding the world as dynamic and ever-changing was, of course, not entirely new in the 19th
century: I would like to mention here the contradictory views of the Ionian natural philosophers like Anaximander and
Herakleitos and between the Eleatian school of Parmenides. Whereas the Ionian natural philosophers stressed the dynamic
nature of the cosmos and were imbued with a keen interest in uncovering its nature - and thus became the precursors of
modern science (Mühlestein 1981: 50-51), Parmenides, a precursor of Plato, believed the moving, changing material world
to be illusory, a mere reflection of a static higher world (Mühlestein 1981: 53).
[5]
I am convinced that, as Winkler (1999: 248) pointed out, the new conception on proto-languages, the family tree and the
comparative method amounts to a recession of linguistics to a pre-19th century level - a reversal of the 19th century
scientific breakthrough. The methodological criticisms of the comparative method, i.e. "back-reconstruction" sharply
criticized by Künnap (1998a: 113), make any uncovering of past historical developments impossible (Winkler 1999: 245).
The new conception of genetic relationships between languages, mainly conflating these with contact relationships, and the
criticism of the language family tree based thereupon makes determination of the genetic identity of a language impossible
and, in fact, meaningless.
On the background of these lies a constant tendency by the "innovationists" to confuse the object of research, language
history, with the methods by which it is researched and the eventual diagram depicting it - e.g., Sutrop (2000b: 198) claims
that something called "the language family tree theory" (2) denies language contacts contribute to "a language family tree
formation" - a notion which was rightly ridiculed by Esa Itkonen (1998a) two years earlier - the genetic identity of a
language is seen as merely a "theoretical construct", the more conservative and stable parts of a language are seen by
Saukkonen (2000: 373) as determining the genetic identity of a language itself, not merely as serving as a tool in determining
these, and so on. This confusion tends to make historical linguistics a meaningless, speculative game in which the actual goal
of any historical linguistics - uncovering real past language changes - is no longer attainable.
THE LANGUAGE FAMILY TREE AND MERGER OF LANGUAGES
The language family tree is a perfect model in depicting that, which it is meant to depict: the break-up of proto-languages
into "daughter" languages, whether this be envisioned as a gradual disintegration process or as a rather sudden event
brought about by the migration of speakers. The language family tree does not depict language contact influences, and it is
doubtful whether this would be desirable at all, since language contact and genetic transmission of language are two
essentially different processes (Winter 1990: 14), and recent attempts to built tree models encompassing both genetic
influence and language contact influence, like for example the model proposed by Raukko and Östman (1995), which I will
treat later, are based upon an unjustified conflation of these two. Of course, a family tree model of languages is not the only
applicable family tree model in historical linguistics. One can draw a family model depicting the development of a certain
proto-phoneme, and here the nodes signify common innovations rather than proto-languages (which are, of course, defined
by a number of common innovations distinguishing them from earlier language stages), as was Budenz' early finno-ugric
family tree. I mention this since the recent critics of the language family tree model in Uralistics have sometimes criticized
language family tree models in a rather eclectic matter rather than the linguistic premises on which these are based (for
example Künnap 1998a: 9-10).
[6]
The family tree model is also criticized (for example in Häkkinen 1984) for presenting an oversimplified view on the rather
murky processes in which dialectal variations of a proto-language emerge as nascent languages, in which isoglosses can
easily transcend later language (and language family) boundaries (Hoenigswald 1990: 443) while not indicating discrete
intermediate proto-languages, and as such is counterposed to the wave model. The two models however differ radically in a
number of important respects: whereas the wave model accounts for cross-language isoglosses in the dimension of space, it
has little time-depth, and likewise, the family-tree model which depicts the movement of languages through time has little
geographical dimension.
It is hard to compare two models which differ even in their objects of research, and I personally see no way in which they
conflict which each other. Although Southworth (1964) tries to reconcile the two models in including cross-branch
isoglosses in a family tree diagram, I believe it is unreasonable to demand that the language family tree model accounts for,
for example, language contact influence and cross-branch isoglosses, and doubt the necessity of such an all-encompassing
model. As Hoenigswald (1990: 442) puts it rather reservedly: "This does not mean that language trees are not useful devices
so long as the impossible - a complete depiction of full history - is not asked of them."
It has been mentioned, for example by Ernst Pulgram (1971: 235-236) and Esa Itkonen (1998: 96-97) that the family tree
diagram is actually implicit in the comparative method itself. As Itkonen (1998a: 96-97) mentions: "Tietysti sukupuun
käsitettä eli kantakielestä lähtevää asteittaista jakautumismallia voidaan käyttää ilman, että piirretään varsinaista
sukupuukuviota." ("Of course we can use the concept of the family tree, a model of gradual division departing from a proto-language, without drawing an actual family tree diagram") . The one premise underlying the comparative method (which always means reconstructing a more or less
uniform proto-language on the basis of present-day linguistic diversity: the reverse would be absurd) and thus the family tree
model is that, while languages often disintegrate into more daughter languages, merger of two (or more) languages into one
single language is regarded as very rare or even categorically denied (Dyen 1987: 107, also Itkonen 1998b: 213) - unlike
disintegration of a proto-language into daughter languages, gradual merger of genetically unrelated languages lacks
empirical evidence, as Laakso (1999b: 60) points out. Thus every serious and relevant critique on the family tree model (and
the comparative method) must proceed by attacking this premise and bringing forth the possibility of language merger. The
most obvious, though perhaps not the only (3), candidates for language merger are pidgin and creole languages, although
these are regarded by for example Thomason and Kaufman (1988: 3, also Thomason 1997: 74) and Dyen (1987: 107) as
languages of non-genetic origin, not the product of normal language transmission from generation to generation. Whereas
Dyen (1987: 107) regards the emergence of creole languages as a historically specific occasion, happening only after the
onset of colonization, Polomé (1982: 246) is rather optimistic about the possibility of discovering ancient creole languages,
although he rejects the possible candidates under examination, like Hittite or Proto-Germanic.
[7]
Also Kaisa Häkkinen, in her famous article Wäre es schon an der Zeit, den Stammbaum zu fallen? (Häkkinen 1984: 1-2),
referred to the possibility of language merger. More recently, Häkkinen expressed her criticism of the language family tree
model in the following words: "Sukulaisuudesta todistava samankaltaisuus voidaan selittää monin tavoin, esim. johtamalla
nämä kaikki kielet yhteisestä kantakielestä tai olettamalla, että maantieteellisesti läheisesti kielimuodot ovat vaikuttaneet
toisiinsa yhtäläistävästi [...] Selitykset eivät sulje pois toisiaan. Mahdollista on sekin, että rakenteeltaan samantyyppiset kielet
ovat spontaanisti kehittyneet samaan suuntaan. Kun mahdollisuuksia on monta, on pelkkää mielivaltaa pukea kehityshistoria
sukupuun muotoon ratkaisua kielitieteellisesti perustelematta." ("It is possible to explain similarities bearing witness of a genetic relationship in many ways, for example by deriving all these languages from a homogenous proto-language or by assuming, that geographically adjacent language forms have converged by mutual influence [...]. The explanations do not exclude one another. It is also possible, that languages of a structurally similar type have spontaneously developed in the same direction. As there are many possibilities, it would be an arbitrary decision to try and fit developmental history in the shape of a family tree without justifying this decision linguistically.") (Häkkinen 1996: 60). I do not agree with Häkkinen - I do
not believe language merger or the possibility of long-term reciprocal contact creating a proto-Sprachbund are valid
alternatives for the language family tree model - but the recent barrage of criticism levelled at the language tree model by the
"innovationists" is essentially different.
Whereas the objections Häkkinen raises could be solved by researching the possibilities of language merger and ancient
creoles, the "innovationist" critics - for example Künnap, Raukko and Östman - reject the concept of languages emerging
from proto-languages out of hand (and with it the results of two hundred years of comparative research), supplanting it with
a methodological sleight of hand in which actual linguistic diversity is projected automatically into past linguistic diversity.
The objections they raise against the family tree model cannot be solved by linguistic inquiry.
RAUKKO, ÖSTMAN AND THE ADVENT OF GENETIC RELATIVISM
The manner in which the "innovationists" have raised the issue of language merger is confusing. Ago Künnap's statement
that "Any Uralic language, just like other languages, is a mixture of languages" (Künnap 1995: 124) can be interpreted in
two mutually irreconcilable ways. According to the first, already mentioned, interpretation, a language is mixed in the
conventional, genetic sense: it has - defining "genetic" in the conventional way - two or more genetic ancestors. Such an,
ontological, conception of a "mixed" language is not in principle irreconcilable with historical linguistics, but it is probably
not this conception which Künnap is aiming at (4). The other, epistemological, conception finds its roots in a 1995 article by
Jarno Raukko and Jan-Ola Östman, ,The 'Pragmareal' Challenge to genetic language tree models. In this article, Raukko and
Östman argue for a far more holistic interpretation of the genetic identity of a language (Raukko and Östman 1995: 42)
taking into account not only the genetic ancestor of a language in the conventional sense, but regarding also languages
which have been in contact with the language in question as "genetic ancestors" (Raukko and Östman 1995: 55, 58), in
which regard they are supported by Künnap (1998c: 607). Thus two essentially different phenomena - language transmission
from generation to generation and language contact influence - are conflated.
[8]
The fundamental mistake of Raukko and Östman's is regarding the genetic ancestor of a language as a purely theoretical
construct: "In the light of our discussion above, Finnish and Swedish may also be said to have a common ancestor. If we
take seriously the all-embracing similarities between Finnish and Swedish, we might even be able to 'reconstruct' some parts
of this common ancestor (although in the present paper we will not attempt to do so)" (Raukko and Östman 1995: 46). The
raison d'être of historical linguistics is to approach by linguistic reconstruction pas language varieties which have really
occurred (Häkkinen 1984: 20), that the reconstruction of a protolanguage is an approximation of a real language variety of
the past (Haas 1966: 130). This presupposes that the existence of an objective, external reality - not necessarily the same
reality as perceived by our senses - is accepted, and that the aim of science and thus also historical linguistics is to attain a
better perception of this objective reality.
This is the sine qua non of historical linguistics, and many of the "innovationists'" mistakes may indeed stem from
misunderstanding or even abandoning this principle (5). The notion that languages do indeed emerge from other languages,
that the aim of reconstruction is to approach an extinct ancestor language, an approach which is made possible by the
comparative, based on the idea that the development of language follows certain laws, the notion that language history and
genetic relationships belong to the realm of objective reality - is fundamentally lacking in Raukko and Östman's conception,
regarding genetic relatedness as a theoretical construct applicable just as well to Finnish and Mordvin as to Finnish and
Swedish.
Also Colin Renfrew, albeit far more cautiously than Künnap et al, entertains the idea of language convergence obliterating
older genetic relationships and forming new language families, thereby referring to Trubetzkoy's article Gedanken über das
Indogermanenproblem (Renfrew 2000: 13, see also Renfrew 1987: 108-109). Renfrew's view, however, does not seem to
be based so much upon a redefinition of the concept of genetic relationship as on a redefinition of the concept of language
family: "[...] the convergence effects operating over so long a period of time may produce a linguistic area or Sprachbund
effect, with similarities developing between the languages wchich are not the product of genetic relatedness but simply of
long-term interaction [...]" (Renfrew 2000: 25). Pauli Saukkonen (2000: 373) also refers to Trubetzkoy while (mis)defining
the concept of genetic relatedness: "The terms 'language family' and 'proto-language' have always belonged together. The
languages of a language family must have some "common" core, a common ancient structure: grammar and word stock. It is
appropriate to ask which the core has to be so that we could speak about the same language family. Trubetzkoy has
enumerated a few qualities that languages have to possess to be regarded as Indo-European ones. If a language has changed
to much that it no more has these qualities, according to Trubetzkoy it does not belong to the Indo-European language
family any longer. This train of thought proves most natural here that language families do not remain the same for ever but
they disperse [?]. We cannot claim, e.g. that Finnish belongs to a 100,000 year old African language family."
[9]
This quote points to a problem in defining the concept of genetic relatedness, which is done in two ways. The first definition
is formulated by Alexander Vovin (1994: 96) in the following way: "[...] two or more languages are genetically related if: 1)
they share a significant number of common vocabulary items and basic morphological markers (if there is any morphology)
established on the basis of regular phonetic correspondences; 2) lexical comparisons include a significant number of
examples with identical semantics.". This definition, thus, requires genetic relationships to be proven (or at least to be
provable) in order to exist, as opposed to unproven or unprovable by the comparative method, and is thus methodological.
It enables us to say, e.g. that Finnish is genetically related to Hungarian but not genetically related ('relatable') to Haida. The
other, precise but also in some situations cumbersome, definition is ontological (presented in, for example, Dyen 1987:
101-102). According to it, two languages are genetically related if they share a common ancestor, irrespective of whether it
was spoken 1,000, 10,000 or 100,000 years ago, of whether the genetic relationship can be proven by comparative
methodology or not. Thus Finnish is certainly genetically related to Hungarian and perhaps also to Haida, and, if one accepts
the hypothesis of the monogenesis of language, probably so. This is a pretty basic distinction, and it is surprising that
Saukkonen led himself astray by it. It is not at all unlikely that Finnish, indeed, belongs to a 100,000 year old African
language family, or an even older one.
It is interesting how Saukkonen's outlook on the genetic relationships between languages - which are historical, diachronic
relations even though genetically related languages are spoken at the same time - is indicated by the current state of affairs in
a language: the "common ancient core", not the often irretrievable threads of language transmission throughout history,
define genetic relatedness. This is indicated in Saukkonen's speculation of how, in the remote future, the large amounts of
Indo-European loanwords will necessitate a reclassification of Finnish and Estonian as "Euro-Languages": "The future
Finnish and Estonian will be Euro-languages, more and more influenced by English and other languages. The names Finnish
and Estonian will be applied to them also for innumerable years to come, but then, however, they obviously do not belong to
the Finno-Ugric language family in any other sense but morphologically. And in a few thousand years comparativists discuss
from where the Finnish and Estonian languages of that time, after their common Proto-European, had obtained forms which
deviate from other daughter-languages as English and German!" (Saukkonen 2000: 373). Little gems of wisdom like these
make me think that, in a far less remote future, the current "breakthrough in present-day Uralistics" will be received with
puzzled amusement by students in History of Linguistics courses, rather like the 17th century hypotheses about Dutch as the
ancestral language of all mankind nowadays.
In her excellent critique of Raukko and Östman's article, Johanna Laakso (1995: 70) points out the absurd consequences of
their redefinition of the concept of genetic relatedness, which is, as apparent in the above, endorsed by Saukkonen as well:
"Is a language - as historical linguists obviously claim - a fixed entity, the essence of which never changes? This does not
mean that even fundamental things could not change, for reasons internal and external. According to an old saying: a spade
is always a spade: sometimes you put in a new shaft, sometimes a new blade, but you still call it "your spade" (and in a
structuralist view, it is the same spade all the time).
[10]
In Östman and Raukko's view, a spade with a new shaft is a new spade. This leads us to a hopelessly complicated situation:
ultimately, we have to consider every person's idiolect (maybe even the different codes s/he uses in different situations) as a
seperate language." Or, taking Raukko and Östman's reasoning to the extreme, Finnish has remained Finnish since the dawn
of time: since (genetic) language transmission is indistinguishable from contact influence, it is impossible to claim that
Finnish did ever emerge from an ancestor language (in the extreme, but logical version of Raukko and Östmans model, it
emerges from an ancestor language every day in everybody's idiolect), thus the Finnish language would have remained the
Finnish language perhaps since the genesis of language itself, only changing continuously due to language contacts.
In the writings of the "innovationists", genetic relatedness is reduced to a mere play of words: "Concretely, we are thus
suggesting that a theoretical construct like 'Baltic Europe' has a similar status to 'Indo-European', 'Germanic', 'Uralic' or
'Baltic Europe'. The status of each of these is highly hypothetical - the main reason for using them at all is methodological -
so that they are equally important or unimportant when we start speculating [sic!] about the 'real ontology' of things. There
is no qualitative difference between 'genetic' relationship and 'contact' relationship." (Raukko and Östman 1995: 59).
The same totally unjustified widening of the concept of genetic relationship is indicated by Künnap: "Ühtlasi on hakatud
kritiseerima keelesuguluse mõistet üldse ja taandama ka seda keelekontaktidele" ("at the same time the concept of language relationship received criticism and began to get applied also to language contacts") (Künnap 1996: 509, see also Künnap 1998b:
420, 1998c: 607). In language, other than biology, acquired traits - the phenotype - are transferred to the genotype, and
there is no intrinsic difference between borrowed elements - the all-embracing similarities between Finnish and Swedish -
which can indeed be more numerous than the indigenious elements, but this does not justify at all equating the transfer of the
genotype - the transmission of language - and the acquirement (possibly but not necessarily due to language contact) of new
features in a language.
The great question which inspires my critique of Saukkonen, Raukko and Östman and Künnap is the following: is the aim of
historical linguistics the uncovering of real language history, which often, in the absence of written records, can only be
researched by examining present-time languages - is the research of language history justified in itself, or is it the aim of
historical linguistics to place present-day languages in a more holistic context, providing them with fancy labels like
"Baltic-European" or "Super-Euro-Finnish"? From the remarks of Saukkonen, Raukko and Östman I conclude that they
totally fail to understand the basic tenets of historical linguistics - first of all, that language history is something which has
really happened, not just determined by the eye of the beholder, and that it can be researched by using certain methods, that
the aim of these methods is to uncover reality and that its results are no mere "theoretical or methodological constructs".
[11]
YOU CAN'T HAVE YOUR PIE AND EAT IT, TOO
The issues raised in the preceding section - the " holistic" view on genetic relatedness, which has no basis in reality, the view
on proto-languages and the language family tree as mere theoretical constructs - points to the relativism which is central in
the "innovationist" conception of historical linguistics, exemplified by Saukkonen's, Raukko's and Östman's regarding of
genetic relationships as a mere label, something which changes through time ("finno-ugric" , "euro-languages" ). The
ahistoricity central to it becomes clear when the "innovationists" (in this case, Künnap and Pusztay) try to go beyond mere
speculation on "alternatives" to the language family tree model, and try to put their views on methodology and historical
linguistics into practice. In a number of publications, Ago Künnap (1998a: 113, 2000: 28) rails against what he calls
"back-reconstruction".
"Back-reconstruction", apparently, means reconstructing a feature in a proto-language and then explaining its disappearance
in one of the daughter languages: "But now a terrible mistake occurs. The traditional Uralists forget that the reconstructed
Proto-Uralic is in both cases only a methodological tool and not a real language (6). They fetishize the Proto-Uralic and
believe it has linguistic evidence. From the pseudo-evidence they begin to back-reconstruct the features and details of the
real living contemporary Uralic languages. The so-called back-reconstruction is the worst they can do. They say: in
Proto-Uralic is the evidence of such and such a feature or detail. They say: if so then all the contemporary Uralic languages
must have the traces of these Proto-Uralic features or details or they have been deleted later. Why must have, why
deleted?!" (Künnap 2000: 28). The same admonishment against what seems to be, essentially, what historical linguists do
appears in an almost programmatical statement co-authored by no less than eighteen other researchers (Amon, Audova,
Balode, Chamberlin, Klesment, Kroll, Künnap, Kuznecov, Kuprina, Mägi, Matisen, Ojamaa, Sarv, Sarv, Saina, Soosaar,
Torn, Toulouze and Verschik 2000: 239). A similar methodological statement is produced by Pusztay (1997: 12):
"Elementtejä, jotka esiintyvät vain joissakin uralilaisessa kielissä, ei saisi - nykyisestä käytännöstä poiketen - johtaa
yhtenäiseen kantakieleen" ("one should not, contrary to current custom, derive elements present in only some Uralic languages from a homogenous proto-language."). It is thus connected to the new conception the of proto-language: is Proto-Uralic was really a
loose Sprachbund of very different languages in contact with each other, common features present in some current Uralic
languages indicate earlier alignments of languages within the Uralic Sprachbund (7), but cannot be generalized to Proto-Uralic
as a whole.
[12]
However, with the methodological critique outlined above, Künnap and Pusztay make itimpossible to distinguish between
features remnant from earlier areal alignments in a rather loose Sprachbund and features which have really disappeared from
some languages and remained in others. In other words, the thesis of a non-unitary proto-Sprachbund is impossible to either
prove or falsify since the alternative thesis - a more or less unitary proto-language, which has not automatically transferred
all of its features to every present-day descendant language, is rejected out of hand. There is thus a circular element in this
reasoning: the non-unitary nature of Proto-Uralic is supported by exactly the linguistic evidence not present in all current
Uralic languages, which is reintepreted in an ad hocfashion to fit a non-unitary view on the protolanguage.
The method named "back-reconstruction" by Künnap is, far from a closet skeleton of the comparative method, part and
parcel of it. As Honti (1975: 125-135) points out (in an article which, like Itkonen 1966, deals with very similar criticism
towards the comparative method), the aim of linguistic reconstruction is not only to clarify the genetic distance between
related languages, but also research which way the proto-language has developed (and diverged) until the present day. A
beautiful example of "back-reconstruction" which would horrify Künnap but which actually demonstrates the power of the
comparative method is mentioned by Haas (1966: 133) (8): the final vowel of some reconstructed proto-Muskogean etyma,
for example *NaNi/u 'fish' and *ixwani/u 'squirrel' is not exactly determined. Comparison with and reconstruction of an
earlier language stage on the basis of related proto-Algonkian, which has *name.kw 'fish' and *anikw 'squirrel', make it
possible to amend the proto-Muskogean reconstructions to *NaNiku and *ixwaniku. As Haas (1966: 133) remarks, this is
"shown as an example in which a new insight in regard to the reconstruction in one protolanguage may be gained through
comparison with another protolanguage."
This seems like the ABC of historical linguistics, and of course it is. Künnap's methodology, however, actually denies the
possibility of features disappearing from languages - the absence of a certain feature in a certain language automatically
indicates a restricted areal spread of that feature in an earlier proto-Sprachbund (or lingua franca, or whatever), and an
earlier alignment of the current languages which retain the feature. Ultimately, the "tree diagram" emerging using such a
methodology would ascribe to Finnish a large number of "genetic ancestors" all sharing a certain feature which would pass
on unchanged to the current Finnish language, an utterly absurd view. As Winkler (1999: 245) pointed out, Künnap's
methodology ultimately leads to the reconstruction of linguistic change becoming impossible - only features already present
in a language may be reconstructed to an earlier phase, even if the case for disappearance of other features is perfectly
supportable, but also leads to an utterly absurd view of the reconstructed proto-language as complete, i.e. exhibiting all
features the "real" proto-language indeed had.
[13]
Both misconceptions occur in Künnap's work. In a number of recent publications (Künnap 1998a: 90-94, 2000b), Ago
Künnap tries to make a case for the new conception of the protolanguage as a loose Sprachbund by trying to enumerate
morphological features supposedly connecting Saami, Balto-Finnic and Samoyed, these features having arisen from a
contact zone at the rim of the White Sea: "At present the zones of occurrence of Nenets and Lapp are quite near to each
other. Why should it have been different thousands of years ago? Why could there not have been, for instance, in the parts of
the White Sea, a contact zone of Lapp, Finnic and Samoyed language forms? " (Künnap 1998a: 90). As it is, Künnap's areal
features are not convincing since the linguistic area they cover is far too wide - feature 3, the genetive antecedent, for
example, totally lacking only in the Ugric languages (Künnap 1998a: 91), feature 4 - the k-imperative - appearing, according
to Künnap, also in Mordvin (in addition to Finnic, Saami and Samoyed), feature 8 likewise. Aside from this, there are also
areal features connecting Samoyed to Ob-Ugric (Künnap 1998a: 71-77).
Apparently only the Permic languages are more or less consistently excluded. Then again, rediscovering the principle of
lateral areas is also an achievement (9). Of course, it is theoretically possible that striking common features demand that
Saami, Finnic and Samoyed be classified as a distinct genetic subgrouping within the Uralic language family, that is, that a
common proto-language would be reconstructed, not a "contact zone" (loan transfer of bound morphemes is extremely
rare). The evidence for such a subgrouping would need to outweigh the evidence connecting Saami and Finnic to the
Volgaic and Permic languages. Simply ignoring the results of comparative research so far and reinterpreting the spread of
features in a language family as remnants of earlier areal alignment will not do.
Interestingly though, Künnap seems to have no problems with "back-reconstruction" if the innovationists can use its results.
One of the putative Uralic substratum features in the Germanic languages, proposed by Wiik and hailed by Künnap (1998a:
104) is the shifting of the accent to the first syllable in the Germanic languages. Now, how do we know that the earlier
accent in the Germanic languages was free? Of course, from comparative evidence of other Indo-European languages a free
accent is reconstructed in proto-Indo-European, which is thought to have become a fixed accent during the seperate life of
the Germanic languages. If the researchers of Indo-European languages had only listened better to Ago Künnap's warnings
against "back-reconstruction", surely they would be of the opinion that the Germanic languages had always had a fixed
accent and moreover, that they had always been Germanic languages who merely came into prolonged contact with other
languages in an "Indo-European" contact zone, proto-Indo-European of course being a figment of the imagination.
[14]
Another contradiction lies in Ago Künnap's attitude towards proto-Uralic. On the one hand, Künnap accuses "traditionalist"
researchers of the following: "They fetishize the Proto-Uralic and believe it has linguistic evidence." (Künnap 2000: 28). On
the other hand, Künnap succeeds in confusing the reconstruction of Proto-Uralic with the actual Proto-Uralic which the
reconstruction is to approach in the following way: "On my part, I have emphasized that if Wiik is right then the traditional
Uralists are reconstructing this very lingua franca as Proto-Uralic because, first of all, in the lingua franca and through it
appeared the common features characteristic of all or at least of the majority of the present-day Uralic languages. At the
same time, there were not so many common features in that lingua franca under observation since it was made best of as an
inadequate auxiliary language. It can be proved by both lexical (according to Juha Janhunen only about 140 word stems
which correspondent to strict phonological criteria, see Janhunen 1981) and grammatical (see Janhunen 1982) insufficiency
in the "Proto-Uralic" under reconstruction." (Künnap 1999: 142).
Probably Ago Künnap is the only Uralist who believes that the fact only 100-something lexical items can with absolute
certainty be reconstructed to Proto-Uralic indicates that Proto-Uralic was indeed a language consisting of only
100-something words (as Janhunen himself recently pointed out, by the way, the small size of the reconstructed lexicon is
due to the structure of the Uralic family tree: any PU lexeme must necessarily be present in Samoyed as well. A "comb
model" would significantly enlarge the number of acceptable PU etymologies (Janhunen 2000: 64)). Among the mistakes of
confusing methods of language-historical research and metaphorical representations of their results with actual language
history, this one certainly deserves first prize. Let's however, for the sake of argument, take the idea of proto-Uralic as a
pidgin language (10) seriously. Janhunen's reconstruction of proto-Uralic morphology hardly corresponds to the radical
morphological loss one would expect in a pidgin language (Mühlhäusler 1986: 152-153), proto-Uralo-Pidgin exhibiting two
different plural markers, a dual marker, at least six noun cases and an elaborate possessive conjugation (Janhunen 1982:
30-32).
The most interesting thing is, however, that Künnap mentions lexical and grammatical insufficiency in proto-Uralo-Pidgin as a proof of its status, but does not concern itself with phonology, since a simplified phoneme inventory with few marked features is typical of a pidgin language (11) (Mühlhäusler 1986: 148-151). The phonology of proto-Uralo-Pidgin is, while far from being a horrorshow like the phoneme inventory of proto-Indo-European, not simple, with eight vowel phonemes and at least sixteen consonant phonemes (not counting marginal consonants) (Janhunen 1982: 23-24). Interestingly, Mühlhäusler (1986: 150) mentions lack of obstruent clusters as a feature typical of pidgin languages.
[15]
Proto-Uralo-Pidgin, however, is extremely fond of obstruent clusters - more than ten percent of the proto-Uralic
reconstructions listed in Sammallahti's (1988) list exhibit them. Now why would such an inadequate, auxiliary language
exhibit a rather complex set of phonemes? The answer is, of course, that phonology is the most closed and semantically
neutral system of a language. Therefore it is easier to reconstruct than morphology - the latter system being more exposed to
loss, expansion, analogical levelling, etc. - and morphemes, other than phonemes, having a semantic dimension which can
provide problems in reconstruction (Haas 1966: 134). So proto-Uralo-Pidgin is off the list and we're back where we started.
The methods by which often superficial (Winkler 1999: 244-245) structural resemblances between some but not all of
related languages are regarded as proof of earlier areal alignments within such a Sprachbund can be regarded as stillborn, as
even Künnap seems unable to use it consistently. The actual applications of these methods to real languages, for example in
the case of the *m-accusative in Balto-Finnic, are quite horrifying. The way Künnap confuses the, necessarily incomplete,
reconstructed Proto-Uralic and the very real language it is intended to approach is absurd in itself, not paying attention to
the obvious inconsistency of using a linguistic reconstruction attained by the very methods which are now frowned upon by
the "innovationists". Worse still however, these results are the nearest the "innovationists" (taking exception to Wiik's
substratum theories, which may or may not be valid but the thrust of which is not incompatible with "traditionalist"
Uralistics) have come in applying the new paradigm to actual linguistics.
Summing up, Künnap (supported by Amon et al.) and Pusztay seem to have made the point that nothing really disappears in
languages. Languages which share a certain feature must be set apart from related languages not sharing this feature in
different areal alignments during a proto-Sprachbund contact zone. This, pure and simply, means that it is impossible to
research language change: the lack of possessive markers in Estonian would be impossible to explain in any other way than
proclaiming Estonian never had any possessive markers - since the presence of possessive markers in an early precursor of
Estonian would necessarily be based on back-reconstruction, similarly, the presence of grade alternation in Balto-Finnic
could not be explained as an innovation since the absence of grade alternation in an earlier language stage can only be based
on the fact that much of the present-day Uralic languages show no trace of grade alternation - thus back-reconstructing the
absence of grade alternation.
These examples are not meant to point out the absurdity of Künnap and Pusztay's reasoning, but to point out in which
manner their conception and "method" is ahistorical - history of course being based on the fact (or theoretical construct?)
that yesterday was a little bit different from today (12).
[16]
PERSPECTIVES
Taking part in the "discussion" on the tree model and the comparative method which has raged in Uralistics the last few
years is something of which the merit is not entirely clear. The critical articles by, for example, Johanna Laakso, Cornelius
Hasselblatt, Esa Itkonen and Eberhard Winkler have not, apparently, dampened the enthousiasm of the "innovationists" in
presenting hypotheses that are almost invariably confused, superficial (13) or totally unhindered by any grasp of historical
linguistics. "The Roots of Peoples and Languages of Northern Eurasia II and III" is the fifth book (and the third collection of
articles) to have appeared in the "innovationist" paradigm and, in my opinion, little progress can be discerned from Pusztay's
1995 monograph, rather the contrary.
It is, then, not suprising that some of the critics of the "innovationists" have participated in the "discussion" rather
reluctantly. In his recent article Winkler (1999: 245) announced that this would remain his only contribution to the
discussion, since so many Uralists had already criticized the ideas of Künnap et al "ohne dass auch nur ein Stück wit
Erkenntnis hätte vermittelt werden konnen". Recently Künnap (2000b: 313) chided Laakso for using a "vulgar and rash
style" (14), presumably referring to an article in which Laakso (1999a: 639) mourned the amount of paper wasted on printing
Künnap's articles. In the same article Künnap, nothing if not imaginitive, compares criticism from the "traditionalist" side
with a patient who is "in a towering rage of pain and hysterical as if being slaughtered by a murderer" (Künnap 2000b: 313).
The more prosaic truth is that the sharp tone of some of the recent criticisms (apart from Laakso 1999a also Häkkinen's
scathing review of "Itämerensuomi - Eurooppalainen maa" (Häkkinen 1999)) may result of being tired of explaining the
elementary principles of comparative linguistics without any discernible result (see also Laakso 1998).
Based on the foregoing, I believe no fruitful "discussion" between "traditionalists" and "innovationists" on the comparative
method - not necessarily the new hypotheses on the place and time of the proto-language - let alone a merger of the two
paradigms or the incorporation of "innovationists" ideas within the traditional paradigm of Uralistics is possible, since the
basic "innovationist" conception of language history is confused and erroneous, the methodology based thereon manifestly
unusable even to the "innovationists" himself.
[17]
The "innovationists" confuse models used to represent research of historical reality with historical reality itself - the language family tree and the reconstructed proto-Uralic - they conflate linguistic concepts - language contacts and genetic language relationships - which should not be conflated, they regard established genetic language families as nice labels which can be applied at convenience while disregarding - or failing to comprehend - the meaning of the concept of genetic relatedness. In short, the "new" conception of the Uralic protolanguage and the language family tree is rooted in misunderstanding the relationships between the object of research, the researcher and the results of research. As the sometimes rather embarassing examples above show, no fruitful results can come from a historical linguistics without historical dimension and from an essentially relativist conception of genetic language relationships. What the critics of the "innovationist" conception have done, rather than "discussing" it, is defending historical linguistics as a scientific discipline.
LITERATURE
Amon, M., Audova, I, Balode, Z., Chamberlin, M., Klesment, P., Kroll, T., Künnap, A., Kuznecov, N., Kuprina, J., Mägi, K., Matisen, E., Ojamaa, T., Sarv, K., Sarv, M., Saina, N., Soosaar, S.-E., Torn, R., Toulouze, E., Verschik, A. 2000: Some aspects and cases of contemporary research of uralic languages and their contacts - Roots II and III p. 239-244.
Beekes, R.S.P. 1995: Comparative Indo-European linguistics. An introduction. John Benjamins publishing company. Amsterdam.
Carpelan, C. 1996: Mikä on alkuperämme? Hiidenkivi 4 p. 10-14.
- 2000: Essay on archaeology and languages in the Western end of the Uralic zone. FU 2000 pars I p. 7-38.
Dyen, I. 1987: Genetic classification in linguistics and biology - Cardona, G., Zide, N.H. (ed.) Festschrift for Henry Hoenigswald on the occasion of his seventieth birthday. Gunter Narr Verlag, Tübingen, p. 101-108.
Engels, F. 1888 [1968]: Ludwig Feuerbach und der Ausgang der klassischen deutschen Philosophie. Dietz Verlag. Berlin.
FU 2000 = Congressus Nonus Internationalis Fenno-Ugristarum 7.-13.8.2000. Tartu.
Haas, M.R. 1966: Historical linguistics and the genetic relationship of languages. Current Trends in Linguistics 3. Theoretical Foundations p. 113-153.
Hoenigswald, H.M. 1990: Language families and subgroupings, tree model and wave theory, and reconstruction of protolanguage - Polomé, E.C. (ed.): Research guide on language change. Trends in Linguistics Studies and Monographs 48. Mouton de Gruyter. Berlin p. 441-454.
Honti, L. 1975: On the question of proto-language reconstruction. Nyelvtudományi közlemények 77 p. 125-135.
Häkkinen, K. 1984: Wäre es schon an der Zeit, den Stammbaum zu fällen? Ural-Altaische Jahrbücher neue Folge p. 1-24.
- 1996: Suomalaisten esihistoria kielitieteen valossa. Tietolipas 147. Suomalaisen Kirjallisuuden Seura Hämeenlinna.
- 1999: Äussere Form und wissenschaftlicher Gehalt. FUF 55 p. 177-183
Itkonen, Erkki 1966: Lappalaisten esihistoria kielitieteen valossa - Suomalais-ugrilaisen kielen- ja historiantutkimuksen alalta. Tietolipas 20. Suomalaisen Kirjallisuuden Seura. Helsinki p. 85-131.
Itkonen, Esa 1998a: Sukupuu ja kontakti. Virittäjä p. 90-103.
-1998b: Keelepuu vajas uuendamist sada aastat tagasi. Keel ja kirjandus 3 p. 213-215.
-1999: There is nothing wrong with the comparative method: part two - Künnap, A. (ed.): Indo-European-Uralic-Siberian linguistic and cultural contacts. Fenno-Ugristica 22. University of Tartu, Division of Uralic languages. Tartu.
Janhunen, J. 1982: On the structure of proto-Uralic. Finnisch-Ugrische Forschungen 44 p. 23-42.
- 2000: Reconstructing Pre-Proto-Uralic typology spanning the millennia of linguistic evolution. FU 2000 pars I p. 59-76.
Julku, K. 2000: Die ältesten Wurzeln der finno-ugrischen Völker im Lichte der heutigen Forschung - Roots II and III p. 125-136.
Kallio, P. 1997: Uralic substrate features in Germanic? Suomalais-Ugrilaisen Seuran Aikakauskirja 87 p. 123-130.
Künnap, A. 1995: The Uralic language tree model and the myth of the migration from the east at the Jyväskylä congress - Minor Uralic languages: grammar and lexis. Tartu and Groningen p. 123-127.
- 1996: Mea Culpa, aga omakeelsed Eesti põlisasukad oleme olnud ehk juba 12.000 aastat. Keel ja Kirjandus 8, p. 505-513.
- 1998a: Breakthrough in present-day Uralistics. Tartu.
- 1998b: Ensiapu koillisesta. Virittäjä 3 p. 420-421.
- 1998c: Olemme väsyneet puihin. Virittäjä 4 p. 607-609.
- 1999: On the oldest Uralic lingua franca - Künnap, A. (ed.): Indo-European - Uralic - Siberian linguistic and cultural contacts. Fenno-Ugristica 22. University of Tartu, Division of Uralic languages. Tartu p. 142-150.
- 2000a: About some morphological features of Proto-Uralic - Roots II and III p. 27-32.
-2000b: Miscellanea Uralica - Roots II and III p. 308-317.
Laakso, J. 1995: A spade is always a spade - comment on "The 'pragmareal' challenge to genetic language tree models" - Suhonen, S. (ed.): Itämerensuomalainen kulttuurialue. Castrenianum Toimitteita 49. Helsinki p. 70-75.
-1998: Tyvestä puuhun noustaan. Virittäjä 4 p. 610-612.
-1999a: Voiko väittää mitä vaan? Virittäjä 4 p. 637-639.
-1999b: Language contact hypotheses and the history of Uralic morphosyntax. Suomalais-Ugrilaisen Seuran Aikakauskirja 88 p. 59-72.
Larsson, L.-G. 1996: Comments on Pekka Sammallahti's paper " Language and Roots" - Congressus Octavus Internationalis Fenno-Ugristarum Jyväskylä 10.-15.8.1995 pars VIII, Jyväskylä p. 34-36.
Makkay, J. 1992: A neolithic model of Indo-European prehistory. Journal of Indo-European Studies 20 p. 193-238.
Mühlestein, H. 1981: Die verhüllten Götter. Neue Genesis der italienischen Renaissance. Verlag Daniel Andres. Biel.
Mühlhäusler, P. 1986: Pidgin & Creole linguistics. Language in Society 11. Basil Blackwell. Oxford.
Polomé, E.C. 1982: Creolization theory and linguistic prehistory - Language, Society and Paleoculture. Stanford University Press. Stanford, p. 237-248.
Pulgram, E. 1972: Family tree, wave theory and dialectology - Keiler, A.R. (ed.): A reader in historical and comparative linguistics. Holt, Rhinehart and Winston, Inc. New York p. 235-241.
Pusztay, J. 1997: Ajatus uralilaisten kansojen ketjumaisesta alkukodista - Julku, K. (ed.): Itämerensuomi - eurooppalainen maa. Oulu p. 9-22.
Raukko, J. and Östman, J.-O. 1995: The 'pragmareal' challenge to genetic language tree models - Suhonen, S. (ed.): Itämerensuomalainen kulttuurialue. Castrenianiumin Toimitteita 49. Helsinki p. 31-69.
Rédei, K. 2000: Urheimat und Grundsprache (Wissenschaftliche Hypothesen und unwissenschaftliche Fehlgriffe). FU 2000 pars I p. 109-126.
Renfrew, C. 1987 [1998]: Archaeology and Language. The puzzle of Indo-European origins. Pimlico. London.
- 2000: At the edge of knowability: towards a prehistory of languages. Cambridge Archaeological Journal 10:1 p. 7-34.
Roots II and III = A. Künnap (ed.): The Roots of Peoples and Languages of Northern Eurasia II and III. Fenno-Ugristica 23. Societatis Historiae Fenno-Ugricae. Tartu 2000.
Sammallahti, P. 1988: Historical phonology of the Uralic languages - Sinor, D. (ed.): The Uralic Languages. Description, History and Foreign Influences. E.J. Brill. Leiden p. 480-554.
-1995: Language and Roots - Congressus Internationalis Fenno-Ugristarum Jyväskylä 10.-15.8.1995 pars I. Jyväskylä p. 143-155.
Saukkonen, P. 2000: On the ancestors of Lapps (Saamis) and Finnic peoples -Roots II and III p. 372-380.
Sims-Williams, P. 1998: Genetics, linguistics and prehistory: thinking big and thinking straight. Antiquity 72 p. 505-527.
Southworth, F.C. 1964: Family-tree diagrams. Language 40-4 p. 557-565.
Sutrop, U. 2000a: The Forest of Finno-Ugric Languages - Roots II and III p. 165-196.
2000b: From the 'Language Family tree' to the 'tangled Web of Languages'. FU 2000 pars I p. 197-219.
Thomason, S.G. 1997: A typology of contact languages - Spears, A.K. and Winford, D. (ed.): The structure and status of pidgins and creoles. Creole language library volume 19. John Benjamins Publishing Company. Amsterdam p. 71-88.
Thomason, S.G. and Kaufman, T. 1988: Language contact, creolization and genetic linguistics. University of California Press. Berkeley.
Vovin, A. 1994: Long-distance relationships, reconstruction methodology and the origins of Japanese. Diachronica XI: 1 p. 95-114.
Wiik, K. 1997: Keelepuu vajab uuendamist. Keel ja kirjandus 12 p. 845-850.
Winkler, E. 1999: Wie kommen die Uralier an den Rhein? Anmerkungen zur neuesten Theorie über die Genese der uralischen Sprachen. Linguistica Uralica 4 p. 241-250.
Winter, W. 1990: The scope of historical and comparative linguistics - Polomé, E.C. (ed.): Research guide on language change. Trends in Linguistics Studies and Monographs 48. Mouton de Gruyter. Berlin p. 11-21
ENDNOTES
1. This is of course true of linguistic reconstruction as well (Haas 1966: 132) and does not disqualify it, nor does it disqualify linguistic paleontology. It constitutes a problem, though, if the location of an ancient homeland is determined on the basis of the geographical location of its present-day daughter languages, e.g. the "centre of gravity" approach (Sims-Williams 1998: 510).
2. There is no such thing as a " family tree theory", i.e. there is no theory which demands that the genetic relationships of languages must be depicted by little black lines on a piece of paper. There are, of course, observations on the evolution of languages and generalizations based thereupon, which can validate or invalidate the language family tree as a useful model in depicting the evolution of languages.
3. Other cases concern languages in which grammar or lexicon has been replaced due to tremendous contact influence, for example Ma'a, a Bantu/Cushitic 'mixed' language. Rédei (2000: 117) recently denied that it would be impossible to genetically classify such languages.
4. It should be granted, though, that this traditional view on the possibility of mixed languages is implicated by Wiik, when he comments on genetic mixing of languages: "Ise ei ole ma niisugust keelupuud leidnud" ("I have not found such a language tree myself")(Wiik 1997: 847). In the same article, Wiik (1997: 846) stresses the fact that protolanguages are real, and not mere theoretical concepts.
5. Compare for example Raukko and Östman's (1995: 59) use of the word "ontology", quoted later.
6. The only one suffering from confusion on this subject is, as I will show later in this chapter, Ago Künnap himself.
7. In this regard it is prudent to remember Erkki Itkonen's warning against this in the case of the putative lexical evidence for hypothetical early connections between the Saami and the Obugrians and Samoyeds: "Metodiselta kannalta on mielivaltaista selittää sellaiset lapin sanat, joilla on sukulaisia vain obinugrilaisissa tai samojedikielissä, rudimenteiksi kaudelta, jolloin lappalaiset eivät muka olleet vielä täysin omaksuneet kantasuomalaisten tai esim. suomalais-volgalaisen kantakansan kieltä. On nim. mahdollista, että kyseiset sanat, jotka kuuluvat ensi sijassa arktisen luonnon sekä peuranpyynnin tai alkeellisen poronhoidon käsitepiireihin, ovat olleet aikoinaan tunnettuja myös lappalaisten lähimpien kielisukulaisten keskuudessa, mutta jääneet pois käytännöstä, kun viimeksimainitut ovat vaihtaneet vanhan metsästyskulttuurin ja siihen kuuluvan kiertelevän elintavan yhä suuremmassa määrin maanviljelykseen ja karjanhoitoon." ("From a methodological perspective it would be arbitrary to regard such Lapp words, which have cognates only in Ob-Ugric and Samoyes languages, as relics from a time when the Lapps would not have totally adoptes the language of the Proto-Finns or, for example, the Finno-Volgaic proto-population. It is namely possible, that the words in question, which belong in the first place to the sphere of arctic nature and reindeer-hunting or of primitive reindeer-raising, were known also among the closer linguistic relatives of the Lapps, but fell into disuse when the latter gradually replaced their old hunting ways and the nomadic life accompanying those with agriculture and cattle-raising." (Erkki Itkonen 1966: 96).
8. I mention this example not only because it is good, but also because the discussion on the language family tree model is not Uralic-specific, but concerns issues universal to historical linguistics.
9. Künnap also lists "Finnic-Lapp-Samoyed supposed common vocabulary" (Künnap 1998a: 93-94). Taking the mean occurrences from a number of etymological dictionaries including those of the Finnish and Hungarian language is a methodological error in itself: not surprisingly, the Finnish etymological dictionary lists 219 Finnish-Samoyed items and 83 Hungarian-Samoyed items (presupposing presence in Finnish), the Hungarian etymological dictionary 182 Hungarian-Samoyed items and 104 Finnish-Samoyed items (presupposing presence in Hungarian). The only way this line of research could be fruitful is basing it on a single, language-neutral source and also researching every other possible combination of languages. I do not want to mention the fact that -by using an etymological dictionary - Künnap uses the results of methods which he specifically denounces.
10. This is the closest approximation to Künnap's idea of an " inadequate auxiliary language" I can think of, a creole language being neither inadequate nor auxiliary. Of course, using traditional linguistic terminology in the right context is something the "innovationists" seem to loathe, regarding their molestation of terms likeSprachbund and lingua franca.
11. To be fair, Thomason (1997: 76-80) provides counterarguments against the general idea of pidgins radically simplified in morphology and phonology, citing Chinook Jargon as a language which has retained some very marked features from its substrate languages.
12. I mentioned them also in the hope the eighteen researchers (in Amon et al 2000) may change their minds. I am not sure whether Künnap will.
13. Eberhard Winkler (1999: 241-242) illustrated the superficiality of Pusztay's and Künnap's search for earlier areal alignments by enumerating some "strikingly similar" features of Mordvin and Swedish. He should be more careful lest these comparisons will be approvingly cited by Künnap in the near future.
14. It seems indeed that Hasselblatt "dumb", "pathetic" and Laakso and Mikone "vulgar and rash style" (Künnap 2000b: 312-313) are no longer eligible for "first aid from the northeast".