The Indus Civilization

by Gregory L. Possehl

  The Language Family of the Indus Writing System

  Many researchers in their attempts to read the Indus script have had to make an assumption on the language affinity of the writing system. Other researchers (e.g., the Soviet team) claim to have discovered this affiliation as a part of their conclusions; they believe it to have been a form of early Dravidian. From the beginning of the work on the Indus Civilization, it has been assumed by the most credible researchers that there were people speaking more than one language in the vast area covered by Indus remains. Marshall stated: “Of the language of these texts little more can be said at the present than that there is no reason for connecting it in any way with Sanskrit. . . . Possibly, one or other of them (if, as seems likely, there was more than one) was Dravidic.”44

  Was the Indus script a rendering of one, or more than one, language into written form? Cuneiform was certainly used to write more than one language, and languages in different language families, just as the Roman alphabet is used today. There is no certain answer to this question for the Indus materials. Some internal consistencies of sign placement and combination suggest that it may be one language, but there are exceptions to these patterns that could be interpreted as a second language.

  Since an assumption of two languages, possibly from entirely different language families, would vastly complicate the research methodologies of some scholars who are attempting to bring order to this body of data, they have proceeded on the assumption that the Indus script renders one language into written form. This is not a proven fact and is a potential source of error in readings attempted so far.

  Another structural feature of the Indus script that is agreed upon is that it admits suffixes but not prefixes or infixes. This feature of the writing powerfully suggests that the parent language was not one of the Indo-European tongues, Sanskrit, in particular. Dravidian languages are possible, but so too are those of the Altaic family.

  No one can draw a linguistic map of the northwestern portion of the Subcontinent in the third and fourth millennia B.C., but there are studies of the larger linguistic scene that help archaeologists to deal with these complexities. For example, D. McAlpin has proposed the existence of a “Proto-Elamo-Dravidian” language family.45 It would have been a set of living languages in the second and third millennia B.C., covering a vast stretch of land possibly extending from southeastern Iran across the southern portion of the plateau into Pakistan and northern ancient India. McAlpin hypothesized that it was languages of this affiliation, not Dravidian itself, that were in the Indus region at the time the Indus script was developed and used. The working hypothesis that many researchers have adopted, which is partly backstopped by the Soviet findings and the structure of “fixes,” is that the script renders one of the Dravidian, or Proto-Elamo-Dravidian, languages into written form.

  THE STATE OF RESEARCH ON THE INDUS SCRIPT

  Introduction

  Decipherment is an art; unfortunately, in the case of the Indus Civilization, an art not yet perfected. It takes commitment and courage to engage in the decipherment game. The decipherment of Hittite is a case in point. The Norwegian, Knutson, saw as early as 1902 that Hittite was an Indo-European language. But he was a temperamental man, unable to accept criticism challenging his insight, and this defeated him. But B. Hrozny had the same thought and pushed his ideas, countered his critics, and stayed with his beliefs. His reward was fame and recognition as a brilliant decipherer.

  The lesson learned from this is that it seems to take a particular kind of scholar to succeed in this decipherment business: someone with the courage to stick with his or her convictions, the will to withstand severe criticism, possibly even public ridicule from colleagues. It even seems to require an unwillingness to recant but to “go down with the ship,” a particular kind of strength.

  While these qualities may be ingredients for success, they do not guarantee it. Hrozny was correct with Hittite and took on a decipherment of the Indus script without apparent success. But he stuck to his hypothesis on Indus writing to the end of his life.46 Strong men such as W. Fairservis, S. R. Rao, A. Parpola and his team, and Knorozov and his Soviet colleagues must be credited with this kind of self-confidence. They have the courage to venture into difficult and “dangerous” intellectual worlds and stay with their beliefs. Through the efforts of these scholars, and others like them, there is agreement on some issues dealing with this writing system, but they fall short of decipherment.

  Some Common Ground

  K. Zvelebil offers some thoughts on points about the Indus script that are (or ought to be) agreed upon.47 These ideas, combined with my own distillation, are as follows:

  The script is to be read from right to left (seals, left to right, but their impressions right to left).

  The system admits occasional uses of left-to-right readings as well as the boustrophedon form.

  The script makes extensive use of suffixes, but lacks prefixes, infixes, and inflectional endings.

  The script is best seen as a form of logosyllabic writing.

  Zvelebil reiterated these points of convergence and added the following:48

  The total number of signs tells us that the Indus script is not alphabetic or quasi-alphabetic.

  The Indus script is not closely related to other writing systems of the second and third millennia B.C.; although some convergence might be found with Proto-Elamite.

  The Indus script is not related to any later Indian script. This includes Brahmi and Kharosthi.

  None of the grids have proved powerful enough to lead to decipherment.

  The Indus script is not likely to have been a written form of an Indo-European language since it apparently lacks prefixes, infixes, and inflectional endings.

  The Indus script is not likely to be a written form of a West Asiatic language since the attribute is placed after the substantive head of the attributive phrase in them. However, there may be a distant, as yet poorly understood, relationship to Elamite.

  The common supposition that the common ending signs of the Indus inscriptions (e.g., the “jar” sign and the “arrow” sign) represent grammatical suffixes like case endings has not been confirmed.

  None of the proposed decipherments of the Indus script can be proved to be true.

  This is the common ground. When one looks at the decipherment efforts, one finds a great deal of diversity in the readings.

  Lack of Agreement in Reading the Script

  While we have the agreed-upon points, there is much disagreement as well. The language affiliation has been proposed to be Dravidian, Indo-European, Sumerian, Egyptian, even Malayo-Polynesian. The method of the decipherer, when method can be determined, ranges from the use of the rebus principle, to the world of tantric symbols. Comparative analysis has been made with scripts far removed for the Indus Civilization, like that of Easter Island. This work was so far-fetched that it is almost comical. The Easter Island script dates to the eighteenth or nineteenth century A.D. The Indus script dates to circa 2500—1900 B.C. Easter Island is 21,000 kilometers from Mohenjo-daro, across vast oceans.49

  The various attempts at decipherment have yielded significantly different readings of the individual signs:50

  The “jar” sign has been taken to be a genitive case marker or the third-person-singular honorific.

  The “arrow” sign has been taken to mark the nominative case, or the genitive case, or the dative case. Some consider it to be a suffix of the oblique case. For Fairservis, it is the term “be powerful,” which is homophonic with the third-person singular in proto-Dravidian.

  The human stick figure is thought to mean simply “man.” Another interpretation is that it is a determinative for “man,” or a word forming the suffix for masculine gender. The Finns consider it a marker for masculine gender. For Fairservis, it is “rule” or “ruler.”

  The “comb” sign has been called “from,” as well as a marker for feminine gender, or still another use as marker of the dative case. For Fairs
ervis, it is, among other things, “his mark” or “belongs to,” if it is a terminal sign.

  The “porter” sign has been called a plural marker and even a separate word interpreted as “god-saver” or “defense.” For Fairservis, it is “watchman.”

  This diversity of thought on the meaning of these and other signs would be vastly expanded if every reading were included. But this is enough to demonstrate that there is no agreement among the various researchers as to the meaning of individual signs, so there should be no surprise that longer messages on the seals have been read in different ways. For example, the message on the “Proto-Siva” or Mahayogi seal has been read by different scholars as follows:51

  Bedrich Hrozny: “Here is the tribute offered to the god Kueya.”

  Swami Sankarananda: “The (aquatic) birds have covered all the waterways.”

  B. M. Barua: Barua believes the message is in Sanskrit and translates into English as “the mountain-worshipped one.”

  Finnish team: “Man of the Star (Siva), the lord of. . .”

  B. Chakravorty: “Satta Kosika with kosika being found in Dictionary of Pali Proper Names.”

  B. Priyanka:

  The seal transliteration is: ha-sa-hi-ma-ma-tra, read from right to left. This suggests that the Indus words sounded something like: trama mahis_’a-ha (left to right) which has a Sanskrit equivalent in tri Maha_ mahis_.a. Mahis_.a has appeared in connection with Krishna Rao’s decipherment. The word means “buffalo” in Sanskrit and would fit with the headdress of the central figure on the Mahayogi seal. Tra for “three” would then refer to the postulated three faces of the Proto-Siva and Maha_ would mean “great.”

  W. A. Fairservis: “An-il the Ruler, He (who) Gathers the Assembled Clans.”

  FUTURE WORK AND THE NEED FOR A TEST OF DECIPHERMENT

  It is possible to identify some areas of research that need the immediate attention of those interested and qualified to undertake original research on the Indus writing system.

  Some Problems That Need Attention: The Working Hypotheses

  In addition to the points of broad agreement already noted about this writing system, several working hypotheses have been formulated that could be useful to those attempting to further our understanding of the Indus script.

  Proto-Dravidian remains the best working assumption for the language affiliation of the script. Broadly conceived, this could include McAlpin’s Proto-Elamo-Dravidian. 52

  Other language families, perhaps even Munda, can be seen as vying for the current position of Proto-Dravidian as the language for the best working hypothesis. Work on this matter is encouraged since the “Dravidian hypothesis” is not proved, and such research will tend to stimulate thought.

  The apparent lack of prefixes, infixes, and inflectional endings in the Indus script rules out Sanskrit as the language of the Indus writing system.

  The rebus principle is a potentially “powerful tool” for deciphering the Indus script.53

  The “additions” thought by some to constitute meaningless allographs are in fact functional elements, perhaps phonetic indicators.

  Determinatives were employed.

  Ligatures occur systematically.

  The structural analytical approach of I. Mahadevan has considerable methodological merit and is largely independent of decipherments employing the rebus principle. Both of these approaches should therefore be employed in parallel, seeking places where the results from each method converge to form a single conclusion.54

  This series of working hypotheses needs to be confirmed by independent work. The confirmation process should form part of the normal science of day-to-day research on the script.55

  The Need for an Agreement on a Sign List

  There is no general agreement on the number of signs in the Indus script. Virtually all of the people who have studied this writing system agree that the sign counts and characteristics place it within the range of a logosyllabic system. There are too many signs for an alphabet and too few for a logographic system like Chinese. But there is disagreement concerning a number of important variables. What are the primary or core signs? What are the variants of these core signs? And which signs are ligatures (combinations) of signs? What constitutes scribal and stylistic variation within the signs? Is there change in these variables over time and between places like Mohenjo-daro and Harappa?

  Dealing systematically with these questions is important for a number of reasons. If answers can be found and independently confirmed (tested), decipherers will have authoritative data to begin making sign identifications based on the shape of the signs, that is, their pictographic qualities. It stands to reason that once the primary signs are identified, then combinations of them in the form of ligatures could be more easily determined and deciphered. It also would be easier to see additions to signs for what they are, either as elements with meaning or as scribal variations or stylistic variants. As the situation now stands, everyone is guessing about the set of primary signs and making claims about sign variation that may or may not be true. From an epistemological point of view, these claims are nothing but assertions and therefore baseless speculation.

  There is one possible exception to this critical observation. When Mahadevan was working on the problem of “sign variation” in connection with his concordance, he attempted to distinguish “sign” from “variant” based on the statistical and positional behavior of signs. Those that looked alike and had similar “behavioral” characteristics within inscriptions were deemed to be variants. Signs that looked alike but had different “behavioral” characteristics were determined to be different signs. This is a beginning, but the undertaking is far from over. More work has to be done to reach a satisfactory conclusion, as seen from the fact that he disagrees with Fairservis on a number of cases.56

  While the Indus writing system has pictographs, this observation does not necessarily help in relating signs to known objects. For example, the “fish” sign was said by Fairservis to be a “knot.” Maybe Fairservis is right; maybe he’s not. Who is to say? Where is the test?

  What about the identifications of the other signs? Is the “arrow” sign an “arrow” or a “spear” or a staff with a triangle on top? What is the relationship between the various “arrow” signs ? These questions, and many others, arise when we try to identify the Indus signs as an initial step to decipherment. What is needed is a detailed study of individual signs that is published, critiqued, and open to revision. With this kind of iteration, conclusions might be reached that would provide confirmation and broad acceptance among the scholars with the special knowledge and skills to address decipherment in a professional way.

  The Need to Study the Writing System in Its Various Contexts

  Fairservis has been the most convincing voice for studying the Indus script within the context of the Indus Civilization. This is intuitively obvious to me, and I tend to neglect its importance when addressing nonarchaeologists, but it is of paramount importance. We must understand the script within the larger context of what we know of the Indus Civilization. This is a matter of theory and of practice. For example, “Fairservis pleads for a reconstruction by the Dravidianists of the obvious artifactual vocabulary familiar to the archaeologist which would include words for characteristically Indus objects (he offers such a word list)—another idea worth attention.”57 In fact, this contextualization of the script is the strongest element in Fairservis’s work with the writing system.58 These skills are not matched by many other workers, Lal and Pande being notable exceptions.

  The Indus script was used in diverse cultural contexts from square stamp seals to signs scribbled on potsherds. It is found in the Greater Indus region at great city sites and in small villages on the frontiers of the Indus Civilization, as well as in the Arabian Gulf and Mesopotamia. There are two primary dimensions to this diversity: (1) the kinds of objects that were inscribed, and (2) the geographical contexts of these finds. It seems highly likely that the message on the typical I
ndus stamp seal with the unicorn and other devices is different from that on the copper tablets, Indus miniatures, and copper tools. If this is the case, then the statistical study of sign position and count should take cognizance of it. The decipherment efforts should not be directed at the script as a general, undifferentiated writing system, but as one delivering messages within the context of different media. The occurrence of Indus writing in the Gulf and Mesopotamia, in contexts whose writing systems are understood, presents special opportunities in terms of deciphering messages.

  The kind of work I have suggested has already been tried, but in a very limited set of examples. Pande’s research on the copper tablets addressed a well-defined body of glyptic material. His more speculative paper on the evolution of an individual sign within the script is another example of the sort of intensive, small-scale research that needs to be done.59 Another example is Franke-Vogt’s excellent study of stoneware bangles from Mohenjo-daro.60 These hard-fired objects are often inscribed and might carry one or more of a variety of messages (owner’s identification, maker’s identification, place of manufacture, etc.). The restricted contexts for writing here, and the number of inscribed stoneware bangles, gives promise to research on these objects.

  The Need for an Independent Test of Proposed Decipherments

  The question of how to test the legitimacy of a claim for decipherment remains a significant one. The appearance of an exemplar with a substantial bilingual inscription, one in Indus script, the other in a writing system that could be read (at the moment, not Proto-Elamite), would settle this. Failing that, the tests have to fall back on how well the proposed decipherment and statistical data available on the script fit together. In the absence of a good bilingual text, the test of decipherment might have to rely on meaningful and consistent patterns; unfortunately, these are concepts that elude precise definition.