Munda in Early NW India

Miguel Carrasquer Vidal mcv at
Tue Apr 3 07:41:14 UTC 2001

On Fri, 30 Mar 2001 07:37:10 -0000, "Gabor Sandi"
<g_sandi at> wrote:

>The main question, however, in my opinion is the original nature of the
>voiced aspirates in IE languages. When IE was introduced into the
>subcontinent, it would have had to have a set of stops corresponding to the
>Brugmannian set {bh dh gh gwh}, distinct from both {b d g gw} and {p t k kw}
>(forgetting about the palatals {k^ etc.} for now). Some linguists do not
>like to postulate this voiced aspirate set, mostly on typological grounds
>(absence or rarity of the corresponding voiceless set). The question is:
>what other reasonable reconstructions can we make? The main authorities on
>Greek (Sturtevant and Sihler are the ones I am most familiar with) all agree
>that Classical Greek phi, theta and chi represent the voiceless aspirates
>/ph th kh/, so simple devoicing of voiced aspirated stops can certainly
>account for them. Outside of India there is no evidence of aspiration
>elsewhere, but Italic /f f h/ can certainly be derived from the changes *bh
> > *ph > *f etc. Germanic, Celtic and Balto-Slavic all show lack of
>aspiration, easily explainable with the changes bh > b etc. (although
>Germanic did not merge the unaspirated and aspirated voiced stops, for it
>devoiced the unaspirated set first).

>To resume: if there were no voiced aspirates in PIE, then what? Voiced
>fricatives  (IPA beta, delta, gamma) are one possibility, but if there were
>no corresponding voiceless fricatives (IPA phi, theta, chi) at the same
>time, the same typological objections can be made as for {bh dh gh}. Other
>suggestions, rather outlandish in my opinion, have been made, but they all
>depend on postulated changes in later IE that are hard to justify (e.g. *b >
>*bh). So, in the end, Occam's razor leads us to keep the PIE set {bh dh gh

>Any comments?

The main objection against /dh/ etc. is not their relative rarity in
the world's languages, but the fact that it's even rarer (or
impossible) to find them in a language that lacks voiceless aspirates
(/th/ etc.).  In the Brugmannian system, as well as in Sanskrit, such
voiceless aspirates did occur.  However, once the voiceless aspirates
were demoted to clusters of voiceless stop + laryngeal (*tH or *Ht),
the position of */dh/ etc. became precarious.  Gamkrelidze's version
of the "glottalic theory" combines this with the (near-)absence of the
PIE phoneme */b/, to arrive at *t = [t] ~ [th], *dh = [d] ~ [dh], *d =
[t'].  However, Gamkrelidze's proposals are untenable in a number of
respects (e.g., there is no trace of confusion between *t and *dh in
Latin or Greek, as one would have expected when *dh > /th/).  The only
reasonable solution that I see is that PIE had no voiced consonants at
all (as indeed suggested by two of the most archaic members Hittite
and Tocharian). That would make *dh [th], and *d [t] (or,
equivalently, *dh = [t] and *d = [t']: in a binary opposition, the
unaspirated member can become polarized to ejective or the
non-ejective member can become polarized to aspirated).  If so, then
*t must have been [tt] (and in Hittite is indeed spelled that way
medially), a fortis (geminate) stop, unaspirated and unglottalized.
The replacement of fortis/lenis by voiceless/voiced would have
resulted in an unstable system *t = [t], *dh = [d:] (murmured) and *d
= [`d] (preglottalized).  If, at the time, the clusters of
stop+laryngeal were already acquiring phonemic status (as in the
Indo-Iranian area), a four way system as found in Sanskrit could
become possible: [t(')] ~ [th] / [(')d] ~ [dh].  The three other
alternatives would be (1) to lose the distinction between murmured and
preglottalized in the voiced series (resulting in [t] ~ [d]), as in
the case of Balto-Slavic (although *d causes lengthening of the
preceding vowel, and *dh doesn't); (2) fortition of the aspirated
member, resulting in Greek/pre-Italic *t => [t], *dh => [th], *d =>
[d]; and (3) fortition of the unaspirated member, resulting in
Armenian/pre-Germanic *t => [th], *dh => [d], *d => [t(')].

Miguel Carrasquer Vidal
mcv at

More information about the Indo-european mailing list