Thursday, February 5, 2009

Possible revised paradigm of the verb 'to be' in IE

Here is possible paradigm of the verb 'to be' in PIE; I wonder if I can account for all the reflexes in the daughter languages.



You will notice that there is no intial *h1; indeed, this is the point. Now then, on to some reflexes.

This paradigm accounts for the Latin forms quite nicely; *esi > es; *esti > est. For the 3rd person plural, Palmer's argument (see this post below) still holds; *s-enti > *s-onti > sunt. Then, the first person plural was created by analogy to the third person plural *smes > sumus and, from this, the first person singular was spawned by analogy *esmi > sum. The second person plural was created by analogy to 2nd person singular form. Palmer's account of the development of the paradigm in Latin remains unchanged.

The Sanskrit reflexes can also develop from these roots with ease.

*esmi > asmi
*esi > asi
*esti > asti
*smes > smas
*ste > stha
*senti > santi

No problems here. Same thing with the Germanic forms; in Gothic and Old English, for example, the plural forms lack an initial vowel.

Given this new paradigm, the odd forms out seem to be the Greek and Anatolian forms, which show an inital 'e' in the plural. This, I believe, can be accounted for through analogy; they came about through analogy to the singular. This is not as strange as it may sound. This exact thing (albeit with an 'a' instead of an 'e') happened in the move from Sanskrit to Pali. Cf.:

Skt. asmi > Pali amhi 'I am'
Skt. asi > Pali asi 'you are'
Skt. asti > Pali atthi 'he/she/it is'
Skt. smah > Pali amha 'we are'
Skt. stha > Pali attha 'you (pl.) are'

The third person plural remained without the initial 'a'.

I have yet to see if I can account for reflexes in Balto-Slavic from the new paradigm. More, perhaps, on this later.


einzelsprachlich said...

Interesting. Wasn't aware of the Pali development.

Now to play devil's advocate a bit. :)

It's a bit odd that a Greek verb that preserves the recessive athematic endings and the enclisis of the late PIE finite active verb would level out the old ablaut pattern (cf. φημί~φαμέν also) when even synchronically abominable-looking verbs like ἵημι~ἵεμεν maintained a strong stem ~ weak stem alternation.

Unlike in Latin, "*esmi" is not totally isolated in Greek—it has a whole morphological class supporting its formal oddity. You might in fact expect analogy to work the opposite way and produce ablaut when none was previously there. Note that some form of ablaut productivity is implicit in the productive Greek perfect, quite a late formation: βέβηκα ~ βέβαμεν is a post-Mycenaean formation from *gʷem- (otherwise **δέβαμεν vel sim.).

"*esmi" is distinguished from the others by its especially high frequency of use, but that's not much of a reason to level out ablaut differences when it's high frequency that preserves irregularities in the first place.

To my mind, I'm saying, the invariant ε~ε looks somehow old.

Now, as for the Anatolian forms, these can't have leveled the ablaut in favor of the singular: they have an e~a alternation. You can argue, perhaps, that these have abolished an old *eš-zi~*š-anzi alternation in favor of the e~a ablaut of root *CeC- verbs like šeš-zi ~ šaš-anzi... I think that is in fact what Kloekhorst says most people in fact do to explain the discomfiting #eC ~ #aC pattern.

Ben said...


OK, it seems that you think the Anatolian forms can be accounted for given the root without *h1. This is good.

As for your objections raised for the Greek forms, I'm still finding a way around them (if there even is one). :-)