23 Mar 2007

Allan Bomhard's "Toward Proto-Nostratic"

(The following uses a special font to render linguistic symbols. Please download and install the Code2000 font [zip] if you're having difficulties with the display. A single asterisk (*) represents reconstructed forms; double asterisks (**) mark implausible forms.)

Allan Bomhard is one of many Nostraticists attempting to iron out the theory of the Nostratic proto-language (see my previous blog: What is Nostratic Theory?). Unfortunately, this ironing has been going on for a while and some may feel that a hole as been blazed into the shirt by now. In 1984, Bomhard published Toward Proto-Nostratic in which he explored in particular the possible relationships between Proto-Indo-European (PIE) and Proto-Afro-Asiatic (PAA).

A daunting task. For me this rings alarmbells in my head immediately because I can tell based on what I know of comparative linguistics that any such relationship is fantastically remote at best. Couldn't we start with something easier? What about the relationship between Indo-European and Uralic which many academics can recognize as a clear possibility? It just breaks my heart to see "long-rangers" go too long-range and bite off far more than they can chew.

Bomhard's position as reflected in his other books is that PIE and PAA are seperated by approximately 15,000 years, as old as Nostratic itself. I can buy that. This is no doubt why he focusses on these two language groups in particular - to attempt to approach the mysteries of Proto-Nostratic by some sort of linguistic triangulation. Conjecture like this is okay as long as it doesn't ignore facts. It's a healthy part of science, but only if you can seperate capably in your mind your untested imagination on the one hand from hard facts or thoroughly established theories based on those facts on the other hand. An "idea" is not a "theory" and a "theory" is not a "fact". Also when I read Allan Bomhard or other Nostraticists, I try consciously to not get stuck into absolutive thinking and pre-judge people as either 100% kooky or 100% infallible. To come away with something from whatever book we may read, we must first ponder logically on what these books can and cannot provide us.

First the bad news about this book: If you're looking to Bomhard to give you a thorough reconstruction of the Proto-Nostratic language, keep looking. Sad as I am to say this, an overwhelming majority of his roots can be rejected right off the hop because they carelessly ignore how the protolanguages he deals with are reconstructed by recognized specialists. This is the way that virtually all Nostraticists shoot themselves in the foot - by reinventing the entire field of comparative linguistics for themselves rather than taking the time to study it. Observe the festival of errors:

  • PIE *dəg-am-, *dg-am- 'earth, ground, man' (p.216)

    The agreed-upon PIE reconstruction for 'earth' or 'ground' is *dʰǵʰom- without controversy. No one but Bomhard and Nostraticists would recognize **dəg-am- as a validly formed root. The word he claims to be 'man' is actually reconstructed as a derived stem *dʰǵʰom-h₁on- 'someone of the earth' (hence Old English guma and Latin homo). Without the agentive ending -h₁on-, *dʰǵʰom- cannot possibly mean 'man'. This is shameful ignorance of well-attested Indo-European grammar.

  • PSem *tyawr- 'bull' (p.216)

    Semiticists purposely reconstruct *θawr- with *θ- (pronounced 'th' as in English 'thin') based on uncontroversial data showing either /ʃ/, /θ/ or /t/. The first sound is definitely not palatalized as he suggests without ignoring what is actually historically attested.

  • PIE *ʔya- 'relative pronominal stem' (p.270)

    Again wrong. The word is commonly reconstructed as *yo- (note that Andrew Sihler, Benjamin Fortson, & Jaan Puhvel cite *yo-, despite all these specialists being aware of the controversial alternative *h₁yo-). Indo-European languages do not show us a trace of such an initial laryngeal here aside perhaps from Greek hos (yet note also ios found side-by-side with hos since at least Mycenaean). In Indo-European Language and Culture (2009), p.144, Fortson describes the added laryngeal as incompletely resolved. When postposed, we only find laryngeal-less *-yo- however and so the belief that this relative pronoun must be a thematic extension of the 3ps pronoun *h₁i- is an idle conviction. There's also no trace of such a laryngeal in Uralic's cognates (nb. FU *jo- and Uralic *-ja [nomen agentis]) nor has Bomhard sufficiently demonstrated that word-initial clusters even originate at all from a common Indo-Uralic ancestor. Ad hoc.

    Even beyond this laryngeal controversy, we are assured that a vowel *a in this pronoun is most glaringly irresponsible in any accepted modern theory of PIE proper whose known reflexes keep it quite distinct from *o.

  • PIE *rəkʔ-/*rakʔ- 'to stretch out, straighten, make straight'

    He fails to notice the actual form, reconstructed long after Julius Pokorny, showing initial laryngeal *h₃ to explain the initial o- found in Greek oregein. The root is therefore *h₃reǵ- and once again Bomhard's casual musings are invalid.

The good news is that Bomhard's many books on Nostratic can be read for the text, not the listings. He makes some valid points about the awkwardness of PIE sound system which, for one thing, oddly lacks *b when the principle of "typological markedness" (see Markedness in Phonology) shows us that it should have it. Many other points like these are things he uses to justify his reworking of PIE altogether as you can see above, however I would argue that what he is in effect doing here is confusing issues of Pre-Indo-European with Indo-European itself. If, for example, he were to reconstruct /*a ablaut and a series of ejective stops in Pre-IE rather than IE, he could keep his theory intact while avoiding the otherwise unnecessary war with IE specialists who naturally reject his work.

To sum things up then, we might say that what Nostraticists appear to be are hopeless extreme- generalists, so caught up in the forest that they forget the trees. Specialists on the other hand tend to reject long-range comparative linguistics a priori because they are stuck looking at a single tree without looking at the wide forest. The future will eventually bring these two extremes together and a new brand of Nostraticist will evolve that both respects specialists yet also has the ability to pull the detailed information these specialists provide us into a general big-picture, one that no longer involves ignoring logic.


Post a Comment