Response to Trautmann

Indian Phonology and Indian Morphology

In China it would appear that it was primarily Indian phonology that took hold, more than other areas of the Indian grammatical tradition like morphology and syntax. As for Europe, Indian understanding of phonetics was considerably more advanced already in Pāṇini’s time (fl. fourth century BCE?) than it was to become in Europe until well into the nineteenth century. Europe did have a traditional terminology for phonetics, originating in the grammarians of ancient Greece, but it was crude compared to the Indian terminology, and showed little understanding of the organs of speech and how they were used. One might have expected that the European Orientalists, once exposed to Indian phonetic science, would have seen its advantages and immediately adapted it for their own use. But although they admired it, and surely would have made quicker progress if they had paid more attention to it, there was a certain time delay before European linguistics fully digested Indian phonetics, and reached a comparable level of sophistication. By the time this happened, European comparative grammar was already well off the ground.

I certainly agree with Trautmann that India was not just a “content-provider” in the development of historical linguistics; its contributions were scientific and theoretical. But I will argue that the intellectual breakthroughs that led in Europe to the new discipline of comparative grammar (and eventually, historical linguistics) arose not so much from Indian phonology as from Indian morphology: the methods of morphological analysis in the tradition of Pāṇini.

Pāṇini’s grammar of Sanskrit, embodied in almost 4,000 sūtras or brief formulaic aphorisms, specifies how Sanskrit words and sentences can be derived from a set of basic roots and affixes through the application of morphological rules, often through a lengthy series of successive steps, much like a computer program. The roots and affixes from which a derivation begins are often quite abstract. For example, all the various Sanskrit verb endings that agree in person and number with the verb’s subject, in all moods and tenses, are derived in Pāṇini’s grammar from a single set of eighteen abstract elements, listed in Table 1. There are two sets of nine each: one for parasmaipada or the active voice, and one for ātmanepada, the middle voice.

 
The capital letters in the table (-P, -Ṭ, -Ṅ ) are formal marks that eventually disappear from the derivation: a final -P, for example, triggers the correct assignment of accent in the verb form to which it is added (by Pāṇini’s sūtra 3.1.4), and is subsequently deleted (by sūtra 1.3.9). The -Ṅ in mahiṄ simply marks the end of the list: the abbreviation tiṄ, consisting of the ti- in the first form tiP plus the final -Ṅ in mahiṄ, is used as a cover symbol for the whole set of eighteen endings.[4] In a third-person singular active verb, the abstract ending -tiP in the table above is replaced in the course of a derivation by one of the appropriate endings -ti, -t, -yāt, -tu, -dha, -tā, -syati, -sīt, etc., depending on tense and mood (Vasu 1891–1898, vol. 3, pp. 589–90). A form like bhavati ‘exists’ (3sg, present indicative active) is derived in a series of steps from the root bhū, each step representing the application of a specific rule: bhū-ti > bhū-a-ti > bho-a-ti > bhav-a-ti (Cardona 2000: 115).

By contrast, traditional European grammars of Latin and Greek simply listed the stems and endings used in different situations. In other words, in the terms introduced by Hockett (1954), classical philology used an “item-and-arrangement” model of phonology, while the Indian tradition used an “item-and-process” model. This suggests that the crucial early theoretical contribution of the Indian grammatical tradition to European comparative grammar was not its sophisticated phonology (whose lessons were not absorbed until somewhat later), but rather its item-and-process model of morphological analysis, which derived words from basic roots and affixes through a sequence of formal derivational steps. What was crucial for European Orientalists was not the fact that they learned Sanskrit, but the way that they learned it: from paṇḍits who would have taught them in the Pāṇinian tradition, probably not too differently from the way Xuánzàng and Yìjìng were taught in the seventh century CE.

Charles Wilkins, author of the first Sanskrit grammar in English, describes how he studied Sanskrit with an Indian paṇḍit who guided him through traditional Sanskrit grammars:

There follows a list of the particular grammars Wilkins relied upon, including Pāṇini’s. Wilkins’s own Sanskrit grammar retains many of the Pāṇinian formal devices: for example, the endings for singular active verbs are given as -tiP, -siP, and -miP, where the final -P is the abstract marker mentioned above; Wilkins calls such markers “servile letters” (Wilkins 1808: 125–26).

As a single illustration of how Indian-style morphology paved the way for comparative grammar, I will cite an example used by Georg Curtius (1820–1885) to persuade classical philologists that they needed to pay attention to the new field of comparative grammar (Curtius 1845: 12). Table 2 shows the Latin and Greek words for ‘homestead’ and ‘kind’; in each case, the Latin and Greek forms are cognate to each other:

 
At first glance, the -us / -os endings in vīcus and oîkos seem parallel to those in genus and génos. But while the plurals in -ī and -oi in the words for ‘homestead’ follow an extremely common pattern for masculine nouns in both languages, the plural forms of genus and génos look somewhat mysterious. Moreover, ‘kind’ is neuter, not masculine, in both languages; so are the few other nouns that work the same way, like opus / opera ‘work(s)’ and vulnus / vulnera ‘wound(s)’ in Latin, or télos / télea ‘end(s)’ and épos / épea ‘words’ in Greek.

According to Curtius, before the rise of comparative grammar, European classicists did not understand the difference between the endings in vīcus and oîkos on the one hand and genus and génos on the other. In fact, the -s endings in genus and génos were historically part of the stem, and had nothing to do with the -us / -os endings of second-declension nouns like vīcus and oîkos. Genus and génos represent the bare stem, with no ending added; in both Latin and Greek, the plural forms come from earlier *genes-a, with the regular neuter plural ending -a (and a variant stem form *genes- instead of *genos-). This *genes-a was affected by regular sound changes affecting intervocalic -s-: in Latin, intervocalic -s- changes to -r-, giving genera; in Greek, intervocalic -s- is lost entirely, giving génea (see Table 3).

It was the study of Sanskrit that made this explanation possible. The Sanskrit cognate to genus and génos is the neuter noun janas ‘race, class of beings,’ whose plural is janasa, with the intervocalic -s- still intact. Sanskrit j- regularly corresponds to Latin and Greek g-, and Sanskrit -a- to Latin and Greek -e- and -o-; even the Greek and Sanskrit accents match. So all the forms are completely regular, sound for sound, as shown in Table 3 below:

 
Just knowing the Sanskrit forms might have eventually pointed European classicists in the right direction; but I would argue that it was learning the Indian tradition’s item-and-process model of morphological derivation that led to the recognition of historical derivations like Latin *genes-a > genera and Greek *génes-a > génea. Moreover, Indian grammar further analyzed jánas as a derivation from the root √jan- ‘to be born,’ with a stem suffix -as- which regularly produces abstract neuter nouns from verb roots, like mánas ‘mind’ from √man- ‘to think,’ vácas ‘word’ (precisely cognate to Greek épos) from √vac- ‘speak,’ and so forth. Indian-style morphological analysis made it possible to identify the root gen- ‘give birth to, beget’ in genus and génos and to clarify the connections of these words with Latin gignō ‘give birth to, beget’ (past participle genitus), Greek génesis, etc., and ultimately to English kind.

Thus it seems that Indian morphology, not Indian phonology, was the primary catalyst in the formation of comparative grammar. It is morphology that primarily occupied Franz Bopp (1791–1867) in his important early work (1816, 1820); in his introduction to translated excerpts from Bopp (1816), Winfred P. Lehmann counts as a “shortcoming” of the work “the almost exclusive attention to morphology,” noting also that Jakob Grimm (1785–1863) had a “similar lack of interest for phonology” (Lehmann 1967: 39).

As far as the history of linguistics goes, then, what was special about the late eighteenth century was that British Orientalists in Bengal, with both the motive and the opportunity to study Sanskrit seriously,[7] had Indian scholars at their disposal—at their command, one supposes—who not only taught them Sanskrit, but also taught them a new way (for them) of imagining how language patterns could be analyzed and understood: a view of language not just as a collection of words and lists of paradigms, but as an intricately structured set of expressions defined by derivations from explicit rules, operating on a limited set of bases and affixes. Scholars familiar with Latin and Greek were able to apply the same approach to those languages, and because they happened to be close relatives of Sanskrit, were ultimately able to work out how these daughter languages could have evolved from a common ancestor.

Knowing enough about Sanskrit to see that it was a close relative of Latin and Greek was not enough in itself to get comparative grammar and historical linguistics started. It is now clear that the family relationship itself had been noticed long before Sir William Jones: van Driem (2001: 1039–51) has documented how the Dutchman Marcus Zuerius van Boxhorn (1612–1653) had already worked out the basic membership of the Indo-European family by 1637: he called the family “Scythisch” (Scythian), and declared that the languages of the family must have come “van een ende de selve afcomste” (from one and the same origin). Van Boxhorn included Persian (closely related to Sanskrit) in the family, but, at first, no Indic languages: he later added Sanskrit, after it was called to his attention by his friend and colleague Claudius Salmasius (Claude Saumaise, 1588–1653). At the time, some Europeans were familiar with Persian, but we get a clue of the remoteness of Sanskrit from the fact that Salmasius’s proposal to include it in “Scythisch” was based on the Indic words quoted by the Greek historian Ctesias of Cnidus (fifth century BCE) in his Indica, a rather fantastic account of India which now survives only in fragments. These were apparently the best data Salmasius had available.

How is it, then, that Jones is still ritually invoked as the genitor of comparative grammar and historical linguistics, in spite of these scholars who anticipated him by more than a century? He was not himself the founder of comparative grammar, but he said the right thing at the right place and the right time: a time when colonial rulers saw learning Sanskrit as a priority, and had paṇḍits available to teach them in the traditional way, using the intricate morphological system of Pāṇini. Van Boxhorn and Salmasius had enough information and wit to detect the existence of the Indo-European language family, but unlike the Europeans of Jones’s era, they did not have the benefit of extensive exposure to the Indian grammar, and therefore lacked the tools to carry the investigation of the family further.

Hidden in Plain View

If the Indian theoretical contribution to historical linguistics is hidden now, it was not always so, or at least not to everyone. One can find prominent mentions of Pāṇini and the importance of his ideas throughout the nineteenth century and into the twentieth. The distinguished Russian-born Indologist Otto von Böhtlingk (1815–1904) published an edition of Pāṇini’s Aṣṭādhyāyī in 1839–1840, with German commentary; later, he consciously followed the model of Pāṇini in writing a remarkable grammar of Yakut, a Turkic language of Siberia (Böhtlingk 1851). Max Müller (1823–1900), the prolific scholar who directed the Sacred Books of the East project, says this about Pāṇini:

Leonard Bloomfield (1887–1949), perhaps the central figure in American Structuralist linguistics, is quoted as saying that Pāṇini was his model for descriptive linguistics, and told his student M. B. Emeneau that Pāṇini’s grammar was one of his “bedside books” (Emeneau 1988: 757). Emeneau also comments: “Anyone who has been exposed to Pāṇini will recognize the Pāṇinian-like character of Bloomfield’s style in his Algonquian descriptions and, in fact, in Language [Bloomfield 1933]” (Emeneau 1988: 755). In a review of a small book on Indian grammar, where he discusses Pāṇini at some length, Bloomfield says:

We find similar comments in Bloomfield’s Language (1933), one of the central texts of American linguistics. In describing the origins of European comparative linguistics, he mentions first the lucky circumstance that Sanskrit turned out to be a close relative of Latin and Greek. Then he says:

Yet Bloomfield himself acknowledges that the Indian contribution to modern linguistics is not well known. In addition to the predictable effects of Eurocentrism, there is also the fact that Pāṇini’s sūtras are so concisely stated that without special study, even a competent reader of ordinary Sanskrit can make no sense of them: they resemble low-level computer language.

The hiddenness of the Indian theoretical contribution cannot be blamed entirely on the Orientalists themselves, who were full of praise for Pāṇini, nor on Bloomfield, who saw matters quite clearly. Rather it seems to result from an amnesia of more recent onset: once constituted as a separate subdiscipline, historical linguistics has been too easily satisfied with simple-minded legends (such as the exaggerated role of Sir William Jones), and too little inclined to a critical examination of its own complex past.