The Relativity of Proof
The existence of numerous points of dispute concerning language classification so well-known to all historical linguists has just been alluded to. What it shows is that comparative linguists do not have a well-formulated and generally agreed on notion of what in fact constitutes proof of a hypothesis of relationship.
If one extends one’s view backward from contemporary linguistics to earlier periods the differing notions of what proof is become even more striking. Consider the following statements from an earlier but not remote period regarding the Finno-Ugric family, actually along with Samoyed, a subgroup of Uralic, a family universally accepted.
The first comes from Kai Donner (1901:129), an eminent Finno-Ugricist and the founder of the journal Finnisch-Ugrische Forschungen.
‘Through J. Sajnovics (Demonstratio 1770) and S. Gyarmathi (Affinitas 1799) Strahlenberg’s well-founded assertion regarding the Finno-Ugric (Uigur) group was proven once and for all.’
The reference here is to a work published by Strahlenberg in 1730. Later than Donner, a well-known Finno-Ugricist Ravila (1935:21) stated that:
In 1770 the Hungarian Sajnovics published his famous Demonstratio, in which, using quite modern methods he proved the relationship of Hungarian to Lapp. By this and by the Affinitas of Sámuel Gyarmathi which appeared in 1799 the Finnish-Hungarian relationship was regarded as finally established.
That this view is widely held by Finno-Ugricists up to the present is shown further by a review of an English translation of Gyarmathi written by Z. McRobbie (1986:159) in which Gyarmathi’s achievements are summed up in the following terms:
Sámuel Gyarmathi was the first scholar to analyze all of the Finno-Ugric languages on a broad basis by emphasizing systematic lexical and morphological comparisons. He was able to define degrees of linguistic affinity pointing out that Vogul and Ostyak are the closest relatives to the Hungarian language. And although Gyarmathi did not utilize regularities in sound correspondences he was nevertheless able to establish a number of Finno-Ugric etymologies still valid today.
Gyarmathi’s work of course preceded the major work of Bopp in 1816 generally viewed as the beginning of comparative linguistics in relation to Indo European. It preceded the first reconstruction of PIE by Schleicher in the 1850’s and of course the first formulations by the Neogrammarians of the notion of exceptionless sound laws in the 1870’s. Yet, as we have seen, Gyarmathi’s work has been widely viewed by Finno-Ugricists as having proved the relationship of these languages.
At the other extreme, and very recently, Callaghan (1991:131) in a review of a dictionary by Harvey Pitkin of Wintun, a Penutian language remarks,
There has been a long debate about whether all or part or none (italics mine) of the Penutian hypothesis is valid. A preliminary step to the determination of remote relationships is competent reconstruction of protolanguages at a lowest level.
Presumably, Penutian includes here California Penutian. Except for the extinct Costanoan whose likeness to Miwok is obvious, we have one or more dictionaries and grammars of all the languages Miwok, Yokuts, Wintun, and Maidu. What more evidence do we need to decide the validity of California Penutian at least, which was proposed by Dixon and Kroeber in 1919 more than seventy years before Callaghan’s statement? Moreover, Callaghan (1958:193) herself had stated that a consensus of opinion among the four field workers including herself concerned with these languages was that California Penutian does indeed constitute a linguistic stock. In the very same issue of IJAL there was an article by Pitkin and Shipley, two of the field workers alluded to in the preceding statement, which contained over 400 etymologies of California Penutian with the usual apparatus of sound correspondence and reconstructed forms.
I have purposely counterpoised here two extremes, the ‘prescientific’ work of Gyarmathi on Finno-Ugric widely accepted as ‘proof’ of the validity of that stock, and the evidence apparently accepted by Callaghan in 1958, but rejected by her in 1991 as inadequate since she states that possibly none of Penutian hypothesis is correct.
From this we see what constitutes proof of relationship varies tremendously depending on the particular language group involved and the historical period. There is then no consensus as to how much and what sorts of evidence is sufficient to ‘prove’ relationships. What we actually have at present, seen in encyclopedias and general reference works on linguistics, is a product of historical accident. Objectively, the California Penutian languages are probably closer to each other than Finnish is to Hungarian. If a stock was obvious enough and especially if it became established early enough it was then and generally is now an accepted grouping. If it was proposed later and in an atmosphere in which terms like the ‘rigorousness’ of the comparative method received more and more emphasis it became controversial or was rejected largely, again, for accidental historical reasons, this time changing fashions and the vicissitudes of the Zeitgeist.
