On the Indo-European origins of the retroflexes in Sanskrit.

• Author: Hamp, Eric P.

Sir William Jones' powerful and often-quoted observation-based on his study of phenomena furnished by the "exquisitely refined" Sanskrit language "both in the roots of verbs and in the forms of grammar" - has long epitomized the statement of the mechanics of genetic comparison. I wish to suggest in this happy bicentenary celebrating Sir William's remarkable achievement that this same Sanskrit language provides us with the clearest adequately complex example we know and can trace in its full development of the mechanics of acquisition by a language of a Sprachbund trait or supposedly diffused areal feature.

It should be noticed at the outset that we rely for our entire demonstration on the results of Indo-European genetic comparison that Sir William's insight set in train. Sir William found Sanskrit to be more perfect than the Greek, more copious than the Latin, and more exquisitely refined than either. It is difficult today for us to attach simple values or unique parameters to these characterizations. He also found comparisons with Gothic and Celtic not quite so forcible. Today we recognize that each of these imposes a more complex task of comparison with Sanskrit or with Greek than the last two with one another: Today there is even debate over just how Germanic is to be related on the Indo-European level to Sanskrit and Greek, though an important element in introducing this doubt has been the Hittite-Luwian evidence. It is also recognized now that the relation of Latin is not so simple as it once was thought to be. The relation of Celtic is really much more straightforward in detail, but the unravelling of the numerous and superficially dramatic changes in prehistoric Insular Celtic lay far in the future from Sir William's vantage point. The fact that he added the old Persian only as a conditional afterthought reflects only the fact that it represents just a fragmentary increment to our Indic knowledge for comparative Indo-European purposes. The core of Sir William's observation rests, of course, on the non-accidental correlation which he perceived "both in the roots of verbs and in the forms of grammar."

Jones thus used as criteria for his determination: (a) a segment of the lexicon, presumably involving (intuitively perceived) sound correspondences, defined by a major and grammatically (i.e., also morphologically) intricate form class, the verb; and (b) the morpho-syntactic configuration of the language. This latter must include such facts as the case structure and inflexion of the arguments of a predicate and of adverbial/directional/locational expressions, the concord for number and for a restricted arbitrary set of genders, the interplay of diathesis in verbs, the correlation of moods with clause relations - all facts which are considered basic and diacritic by a Western classicist. These criteria enhanced the specificity of his observation, thereby reducing the possible element of chance and the tautology of human universalism, both factors which unfortunately flaw some of the most enthusiastically heralded work of today.

As genetically related dialects and languages diverge, thus creating for us language families, factors typically intervene to effect this divergence, which must be traced to the effects of social contact. Some, although not all, social contact comes in the form of contacts with other languages, into relations and contiguity with which the linguistic objects of our study have entered. If a language is transported to a new terrain such contacts are nearly inevitable; languages which move into a human void - perhaps Icelandic and Polynesian - should be the only cases exempt from this historical condition.

M. B. Emeneau, in his justly renowned article on "India as a Linguistic Area,"(1) has remarked, following the painstaking work of earlier scholars as well as his own wide field experience, that retroflex consonants are found in most languages of India. Consonants affected include stops, the nasal, in some languages also the sibilants, laterals, tremulants and even others. The linguistic stocks embrace Indo-Aryan, Dravidian, Munda, and Burushaski. So:ra: is stated to lack them, and therefore, according to Burrow, the retroflexes would not be Proto-Munda. Earliest Sanskrit shows them, yet they are certainly not originally Indo-European. However, they are certainly Proto-Dravidian, it is claimed, and "not the result of conditioning circumstances"; south Dravidian is characterized by three distinct points of articulation: dental, alveolar, and retroflex. Therefore, we can safely impute retroflexes to...