A Property of "Grassmann's Law" in Indic

IN INDIC, original voiced aspirates (IE bh dh gh g (to) h) are dissimilated into unaspirated voiced stops (Skt b d j g) if the next syllable begins with an aspirate, appearing in its ordinary shape (Skt bh dh h gh). Something similar happens in Greek: before an aspirate (Gk ph th Ich, from IE bh, etc.) in the following syllable, an old aspirate goes to the corresponding plain p, t, or k. The two IE subfamilies affected are precisely those in which aspiration functions as a distinctive feature.' Whatever the relation between the two processes may be,2 and whatever may account for the differences between them, the existence of such differences is clear enough. The principal one lies in the outcome itself. In Indic (where both voiced and voiceless aspirates occur) the dissimilation product of, say, dh remains voiced and hence coincides with the old d. In Greek, on the other hand, where all voiced aspirates are phonetically devoiced and where there are no other aspirates, the result is t.3 Thus, Gk. tei'ch(os) 'wall' corresponds to Skt deh(O) 'rampart,' both representing IE dh-'h. This was H. Grassmann's finding.4 It accounts for a good many cases, some in which reduplication is involved (Gk ti-the-si: Skt dd-dh&-ti 'puts') and others in which the inducing and the induced consonant are the final and the initial consonant, respectively, of a root. This seemed sufficient. Questions of phonemic compatibility and canonic morph structure were not raised in a systematic fashion in Grassmann's time. Meanwhile certain severe and highly symmetrical mutual restrictions upon the cooccurrence of consonants in IE roots have become known. The particular restriction which interests us here is that which prohibits voiceless stops from appearing together with aspirates. The type bhewt pewdh does not anciently exist; neither, incidentally, does bewd, i. e. a combination of two plain voiced stops. It is conversely implied that pewt pewd bewt bewdh bhewd bhewdh occur freely.5 This would seem to impart to the Greek variety of Grassmann's law a status quite different from that of its counterpart in India. If IE bhewdh yields Gk peuth (Hom. pe'thomai 'inquire'), this kind of change cannot ever lead to homonymy with another inherited root, since pewdh is excluded as a source. In other words, the dissimilation is here not a step in a merger process but only in a reassignment of phones in a neutralized area. Precisely for this reason, occasional Greek forms like thuphlos for tuphlos 'blind' have little importance. They should certainly not be taken as manifestations of a serious assimilatory trend capriciously counteracting the fundamental dissimilation.7 The fact is rather that aspiration functions in Greek very largely as a property of the (discontinuous) consonant sequence as a whole, its precise location in one segment or another remaining for some time non-distinctive.8 The Indic picture appears at first in a different light. We are given to understand that for a Skt bodh there are indeed two possible sources, namely bhewdh (bodhati 'awakens,' cognate with Gk pezithomai) and tbewdh-a specific item the lack of which is thought to be only a lexical accident. If, however, a search is made for a more appropriate case, none will turn up. No homonyms have arisen. This is perhaps not for want of a basis in the ancestor, since a count in Pokorny 9 (where of course the good is thoroughly mixed with the bad) yields about 38 'roots' of the type gerbh (that is

Epistemology · Political science · Property (philosophy) · Pure mathematics · Law · Linguistic Variation and Morphology · Linguistics and language evolution · Mathematics · Philosophy · Phonetics and Phonology Research

• Aus dem leben eines lautgesetzes

  Sergej A Romaschko•Historiographia Linguistica•2000

  Add to list