With more than 200 million speakers, the 26 Dravidian languages constitute the fifth largest language family of the world. The four great literary languages - Tamil, Malayalam, Kannada and Telugu - are spoken in southern India, and by emigrants worldwide, especially in southeast Asia. Of the remaining 22 mainly tribal languages, Brahui is spoken in Pakistan, and Kurux in India, Nepal and Bangladesh.

It was in 1816 that Francis Whyte Ellis recognised that Tamil, Malayalam, Kannada, Telugu, Tulu, Kodagu and Malto form a distinct "South Indian" language family. Robert Caldwell's pioneering Comparative Grammar of the Dravidian Languages (1856) took "Dravidian" from the Sanskrit dr?vida , which denotes Tamil and its cognates. Caldwell's work (based on 12 languages) remained fundamental until the monumental Dravidian Etymological Dictionary by Thomas Burrow and M. B. Emeneau (1961).

Bhadriraju Krishnamurti's weighty PhD thesis Telugu Verbal Bases (1961) was written in 1955-56 under the guidance of Emeneau. In 1985, Krishnamurti published a grammar of his mother tongue, Telugu, (in collaboration with J.

P. L. Gwynn) and in 1969 a book on Konda or Kubi, a tribal language also spoken in the state of Andhra Pradesh. Both books are descriptive as well as comparative, and throughout his career, Krishnamurti's emphasis has been on the comparative study of the Dravidian family and his research papers introduced innovative ideas (for example, the reconstruction of a proto-Dravidian laryngeal phoneme, which survived in Old Tamil alone).

ÁñÁ«ÊÓƵ

The Dravidian Languages is the crown of Krishnamurti's life work, a superb synthesis of comparative Dravidian linguistics. There is a marked difference between this book and another with exactly the same title, The Dravidian Languages (1998), edited by Sanford B. Steever. The bulk of that book is devoted to the description of ten distinct Dravidian languages, while the comparative study of the language family is practically restricted to Steever's introduction. Krishnamurti's book takes into consideration all 26 languages, and while he by no means neglects description, it is subordinated to comparative study. All comparative statements are supported by nearly exhaustive evidence. Nothing so comprehensive has been published in this field before.

Besides the issues relating to the history of the language family, the book covers phonology, word formation, nominals with their gender, number and declension, the verb, adjectives, adverbs and clitics, and syntax. The chapter on lexicon deals with loanwords from Indo-Aryan, from Perso-Arabic sources and from Portuguese and English, neologisms and structured semantic fields.

ÁñÁ«ÊÓƵ

In presentation, the data from the individual languages are grouped according to the different branches of the family, and reconstructions are given for each developmental stage as well as for proto-Dravidian. One of the major results of Krishnamurti's work is the accumulation of detailed evidence (presented in tables and diagrams with isoglosses showing shared dialectal features) for the revised classification of the Dravidian languages into four groups (north Dr., central Dr., south Dr. I and south Dr. II), which Krishnamurti has been proposing for the past 30 years and is now generally accepted.

The other descendants of pre-Tamil split before the palatalisation rule occurred in early Tamil in about the 3rd century BC. Pre-Tamil, in its turn, should have split off from the rest of South Dravidian I (Tulu and Kannada) in about the 6th century BC. Several Dravidian loanwords in the Rigveda attest to the presence of Dravidian speakers in the Indus Valley around 1500-1300BC, about the time the North Dravidian branch is likely to have started differentiating from other descendants of proto-Dravidian.

"On cultural grounds, I have suggested the probability of the Indus civilisation being proto-Dravidian," states Krishnamurti; but "the Indus seals have not been deciphered as yet", and he has refrained from discussing this topic further. (In my book Deciphering the Indus Script , I have presented systematic interpretations of 25 Indus signs, forming interlocking compounds attested in Dravidian languages and agreeing with contextual clues. Admittedly, the evidence is scanty, yet more than suggestive.) Likewise, Krishnamurti has been unwilling to devote much space to the speculations of distant language relationships. In a short review of all the principal proposals, he concludes that there is no convincing evidence for a genetic relationship between the Dravidian and any other language family.

Readiness to suggest explicit reconstructions has characterised Krishnamurti's work since his PhD. The new book is full of forms with an asterisk, including a pilot list of proto-Dravidian cultural vocabulary.

ÁñÁ«ÊÓƵ

This anticipates his proposed revision of the Dravidian Etymological Dictionary , with reconstructions (lacking in the first two editions) added.

Numerous insights are scattered throughout the book. There is space here for just one example of culture-historical interest. The Greek loanword oryza dating from the 4th century BC, indirectly the source of the English word rice, does not go back to Tamil arici as has been supposed since Caldwell, but to proto-Dravidian * warinci , "rice, paddy".

Krishnamurti's new etymology solves also the old problem concerning the origin of Atharvavedic Sanskrit vr?hi , "rice", and its Iranian cognates, including Khotanese Saka rr?ysu  (<* wr?ze -) "rice", (Zoroastrian) Middle Persian blnc (= brinj ), Modern Persian brinj "rice" and Pashto wr?ze (plur.) "rice". I would suggest that the word was taken over from proto-Dravidian into proto-Indo-Aryan as * wrinjhi in the Swat Valley, where rice cultivation was introduced in the Late Harappan period, around 1800BC.

For any scholar interested in Dravidian languages, this book is indispensable.

ÁñÁ«ÊÓƵ

Asko Parpola is professor of Indology, University of Helsinki.