Re: [tied] Bartholomae's Law and Grassman's Law

Residual Vedic forms such as <daks.i>, <aduks.at> or <dipsati> (desiderative of /dabH-/ 'deceive', thus still in Sakalya's language, while Panini knows only <dHips-> or <dHi:ps->) show that Grassmann's Law preceded deaspiration (except in word-final clusters as in Skt. bHut < *bHudH-s, or aor. (a)dHak < *dHegWH-t) where the deaspiration must be very old, presumably pre-Indo-Iranian), and that forms like <dHaks.i> etc. are secondary. <bHudbHis> appears to be fairly late too (though still pre-Vedic); it shows the application of a new principle (more recent than deaspiration, and violated by the archaic examples above): underlying aspiration in a root ought to surface -- either through Bartholomae's Law, if applicable, (which is why <buddHa-> or <dagdHar-> were not remodelled) or by being "thrown back" to the initial (the synchronic converse of Grassmann's Law), which is the default option. The <bH> of <-bHis> is underlying and prevents the root aspirate from emerging in the suffix, so the only place where it can go is the initial. Interestingly, there are imperatives like <dugdHi> which seem to preserve the old (Grassmannian) form. Imperatives are loosely related to the rest of the paradigm and more likely to resist restructuring.

This does away with the ordering paradoxes; what remains to be determined is the relative ordering of Grassmann's and Bartholomae's Laws, both of them older than non-final deaspiration. And here I've spotted a hole in my own argument: once the paradoxes have been resolved, the ordering of the two is no longer critical. Grassmann's Law affects only the first aspirated stop, while Bartholomae's has to do with the second. They can't get into each other's way, so either ordering will work. We know that Bartholomae's Law is a Proto-Indo-Iranian change and that Grassmann's Law is _at least_ Proto-Indo-Aryan, but since Iranian and Nuristani have merged former plain and aspirated voiced stops, we can't hope to find any evidence for or against Grassmann's Law there. This of course doesn't lead to any new paradoxes; it's just impossible to tell (unless there is some evidence I've forgotten to consider) what the historical ordering was.

Piotr

----- Original Message -----
From: P&G

To: cybalist@yahoogroups.com

Sent: Thursday, April 18, 2002 12:15 PM

Subject: Re: [tied] Bartholomae's Law and Grassman's Law

>>Bartholomae's >> Law and Grassman's Law, ... >could you explain how you understand the two > laws "imply" two different orders of events? I have just read Piotr's magisterial and magnificent posting on this (for which thank you, Piotr!) but I still thought a Sunday School outline of the problem might also help (I should say "apparent problem" in the light of Piotr's posting). You'll find various ways of explaining it in Collinge's "Laws of IE". He quotes Stemberger, who says it is "very messy". In brief: Bartholomae's law involves a deaspiration, and should be thought of as happening in two steps: 1. assimilation: Dh + t- > DhDh even across a morpheme boundary 2. deaspiration: Dh > D when it is the first element in a cluster The deaspiration alone is seen in forms like labdhva < *labh-dhva. Now the trouble is that Grassman's Law appears at times to occur before this deaspiration, and at other times after it. Schindler (1977, reference below) says "No system in which Grassmann's Law applies to both reduplication and initial aspirates can [explain the evidence] without losing one of two generalisations." He says if Grassman's Law comes first, there must be two deaspirations, and if there is one deaspiration, there Grassmann's Law must be split into two. (a) The usual pattern is seen in *bHudH-ta > buddha-. Here the order apears to have been Grassmann, then deaspiration (*bHudH-ta > *budH-ta > *budHdHa- > buddha). [Though the middle two steps could be the other way round.] If the deaspiration in the cluster occurred first, Grassman's Law would no longer apply, and we would expect the outcome to be *bHuddHa. Example after example shows this apparent pattern of Grassman followed by Bartholomae (b) Unfortunately when the deaspiration occurs in other contexts, it is clear that Grassman has not preceded it! e.g. future bhotsyati, aorist abhutsi and compare the noun stem -budh, which gives bHut, budHam, bHudbhis, etc Here Grassmann's law could not have applied before the deaspiration. There have been many attempts to disentangle what's going on. Some folks try to have two deaspirations, one before and one after Grassmann, some try to redefine the morpheme boundary so Grassman would still apply to *bHuddha, some even deny the reality of Grassmann's law at all. One version of the law even allowed for "optional" aspiration. There are also bunches of exceptions to whatever rules we put forward. There was an article by Jochem Schindler in JIES 1977, substantially along the lines Piotr suggests, in which he says that original roots of the form *Dh-Dh end up with four different forms, chosen according to context. So PIE *bHudH appears in Skt as: budh (before a vowel or resonant) bud (before dh-) but (before th-) and bhut (before #) You will have noticed that this does not explain the future forms or the noun forms above. Peter