Friday, September 16, 2011

Indus Script and Semasiography

Seal M-314, bearing the longest Indus inscription: CARTWHEEL / BI-QUOTES //
TRI-FORK / CIRCLED VEE / QUADRUPED (probably 3 units of information).

At this point, we should examine the evidence presented against the hypothesis that the Indus script is a writing system (Farmer, Sproat, and Witzel 2004).  There are several indicators noted in this famous paper, all of them suggesting that this script is not a fully developed writing system:

(1)    Brevity of the inscriptions, with an average of 4 to 5 signs each, the longest by any count containing fewer than 30 signs;

(2)    The absence of evidence that longer inscriptions once existed on a perishable medium (such as remains of “inkpots, brushes, palettes, styli, pens, and other literate paraphernalia; representations of scribes, texts, and writing instruments in art or pictographic scripts; and major changes in the shapes and orientations of signs tied to scribal attempts to increase the efficiency of copying long texts” (2004: 25);

(3)    The odd sign frequencies, described above, in which twenty signs account for over 50% of all sign occurrences, while the vast majority of signs are rare;

(4)    The low frequency of sign repetition within single inscriptions, which suggests that the signs did not represent phonetic information;

(5)    Within the group of repeating signs, some seem to appear in groups (such as the various types of FISH that appear in a series), and when a specific sign does repeat in an inscription, it may repeat several times in a row, which suggests a possible enumerative function rather than encoding of phonetic information;

(6)    Besides the large number of singletons and rare signs, new excavations tend to turn up still more of this class, suggesting that new symbols were created as time went on, a characteristic not normally found in phonetic writing;

(7)    Some non-linguistic symbols “exhibit a kind of linearity that is not dissimilar from the sort found on some (but by no means all) Indus inscriptions,” including VinĨa inscriptions from southeastern Europe and the symbols of deities on kudurru and elsewhere in the Near East;

(8)    Sometimes symbols with seemingly clear meaning behave in odd ways, such as apparent numerals appearing in inscriptions where they appear not to qualify or enumerate anything -(e.g., K-59 which reads: DEE-SLASH / SINGLE POST / STACKED THREE / STACKED FIVE; or K-49: THREE QUOTES / STACKED SEVEN / MAN WITH SHOULDER YOKE).

The conclusion of Farmer et al is that “Indus inscriptions were neither able nor intended to encode detailed ‘messages,’ not even in the approximate ways performed by formal mnemonic systems in other nonliterate societies.  Their most likely function, as suggested by Near Eastern parallels, was to associate individuals, families, clans, offices, cities, festivals, or professions, etc., with specific gods or their celestial correspondents, partly for identification purposes and partly to draw down whatever magic was accessible through those gods’ symbols” (2004: 42-43).
A sealed proto-Elamite economic tablet from Susa, with two numerals;
perhaps to be read "twelve [animals]" (after Potts 1999: 61, Pl. 3.1)

We may wish to examine some of these arguments in greater detail.  For example, some scholars counter the fact of the brevity of inscriptions by presuming that longer texts did occur, but on perishable media.  They may cite the existence of long texts on such media in later, historical periods, as evidence by analogy (e.g., Parpola 2009: 54).  Historical arguments of this type are very weak evidence, since we know of a great many things that only came to exist in later, historical periods.  For example, we cannot look at the bronze artifacts from the Indus Valley and say that, because iron and steel implements also appear later on, the Indus Civilization must also have produced iron and steel – in the Bronze Age.  No, in the Bronze Age, implements were made of bronze, while iron came into use later.  No doubt many objects were made of perishable material during the Bronze Age and so are lost to us.  For example, cloth has not survived from this period in the Indus Valley.  However, unlike the proposed long texts, there is evidence of cloth and clothing from Harappan statues and figurines that were modeled to depict clothing.  Thus, even though the cloth itself is gone now, we do find evidence that it existed during the Bronze Age.  This is the type of evidence that Farmer et al do not find when it comes to longer Indus texts.  Not only are there no long texts, there is no indirect evidence that lengthy inscriptions once existed.
Analysis of proto-Elamite tablet TY 7, showing the three typical parts: heading ("hairy triangle"),
entries in numerical form concerning grain, and total from the reverse (Damerow and Englund 1989: 39). 

This by itself is insufficient to make the case against Indus script being a fully developed writing system, of course.  Most of the evidence for a Lycian script comes from short inscriptions, for example, as is also true of Etruscan (although in each of these cases there is also at least one long text).  It is sometimes argued that absence of evidence is not evidence of absence.  But, of course, when no evidence is forthcoming, that does suggest something.  In other words, absence of evidence is indeed a kind of evidence for absence; it just is not proof. 

Regardless of this point, Farmer et al have additional arguments that are stronger.  The statistics on sign frequency suggest that Indus “script” is a proto-writing system along the lines of proto-cuneiform and proto-Elamite.  They themselves note this parallel.  Now, an Assyriologist may not focus on proto-cuneiform not being a fully developed writing system, because, in time, it did develop into such a system.  Determining the precise point at which this happened seems unimportant because of the continuity. 
A tag from Egyptian Tomb U-j at Abydos bearing possible early hieroglyph
(after O'Connor 2009: 145).

Egyptian hieroglyphs also had an early phase that may have been proto-writing, as evidenced by the brief inscriptions on tags (small, flat, rectangular objects of wood or ivory on which there may be a glyph or two) found in pre-Dynastic tombs at Abydos (O’Connor 2009: 143-147).  Here again, the determination of the exact point at which proto-writing developed into true writing is not a major focus of scholars.  That it did develop into true writing is clear.

But in the case of both proto-Elamite and the Indus script, there is no continuity with later writing systems.  Instead, these early systems died out.  This is also the case with Aztec proto-writing in Mesoamerica.  Unlike the neighboring Mayans, the Aztecs (and their predecessors) did not develop true writing and eventually their symbols ceased to be used at all.
Detail from a Mixtec "map" (lienzo) with year sign (AO combination), attached numeral indicating date,
footprints on path indicating movement, and a geographical glyph for Zacatepec (Smith 1973: 268, fig. 89).

Moving to the second argument against the Indus script thesis, Farmer et al note the absence of Harappan artwork depicting scribes at work along with the equal absence of archeological evidence that writing occurred.  This, too, is rather weak evidence, but evidence it still is.  The third argument, the peculiar frequency data, with many rare signs versus a small core of frequent signs, was discussed in the previous post.

The fourth argument focuses on the behavior of the frequent signs, the core symbols that appear again and again.  I noted in the last post that the Egyptian glyphs representing single consonants tend to occur the most frequently, even in the limited area of royal names and titles in cartouches.  I noted that the bread loaf representing t is one of the most common glyphs, a symbol that appears twice in King Tut’s name (in the New Kingdom) and twice in the name of two First Dynasty kings, Atet I and II (during the Old Kingdom).  Other glyphs are sometimes doubled, as is the case with the irrigation ditches (N24) representing Semti, fifth pharaoh of the First Dynasty, and as is the case with Bebi, a king of the Second Dynasty (the glyph of a human leg and foot, D58, is doubled, representing b).  The final element in these names is –i or y, a sound that represents dual number in the Egyptian language.  In the names, the same sound may or may not represent the dual.  Glyphs can also be tripled to represent the plural, as in the name of the fifth king of the Fourth Dynasty, Menkaure.  But besides such consecutive repetitions, glyphs can be repeated in non-adjacent positions.  This is the case of the chick (G43) representing w in the name of Khufu (Fourth Dynasty), as is also true of the reed (M17) in Asa (son-of-Ra name of Shepseskara of the Fifth Dynasty).
Tablet H-764AB with inscriptions: CIRCLE / 3 TOED FOOT / POT (A side);
TRIPLE CUPS (B side) (tripling may be enumerative, i.e., equal to 3 + CUP).

A fair number of Indus signs appear either singly or doubled, e.g., the GRID, the CEE, and CIRCLE.  Less commonly, one finds a symbol repeated in a non-consecutive position: M-396 contains two of the BLANKET WITH FOUR TICKS separated by four signs; H-369 contains two of the POTTED ONE separated by one sign; M-634 contains three of the CARTWHEEL with one sign separating each instance.  If this type of repetition were common in Indus inscriptions, this would suggest that phonetic information was conveyed by the common, repeating signs, as in Egyptian.  But since this kind of repetition is infrequent, the argument against phonetic encoding is supported.
Reverse of seal M-326B with inscription (from right): STRIPED TOP / EX / THREE QUOTES / CARWHEEL // BARBELL ON POST / TWO POSTS & BACKSLASH / BARBELL ON POST / (?) POSTS (corner abraded).
(If we add these signs, those in the inscription on the A side, and those on the edges, there are 20 signs altogether).

The fifth argument addresses the kind of sign repetition that occurs more often, doubling and tripling in consecutive series.  Parpola mentions this:

A few signs occur with three- or fourfold repetition: [TRIPLE BOATS, TRIPLE RECTANGLES, TRIPLE CUPS, QUADRUPLE TRI-FORKS].  Of these signs, only [CUP] and [TRI-FORK] are preceded (in other texts) by groups of strokes, which evidently denote numbers.  The sequence [CUP / CUP / CUP] occurs on the reverse sides of some of the miniature tablets from Harappa, which normally show the sequences [CUP], [TWO POSTS / CUP], [THREE POSTS / CUP], or [FOUR POSTS / CUP].  The sequence [CUP / MAN HOLDING CUP] in the broken seal inscription 1165 [CUP / MAN HOLDING CUP / BI-QUOTES // BELTED FISH /...] is paralleled by the sequence [TWO POSTS / MAN HOLDING CUP] on the reverse of tablet H-247.  The sequences [TRIPLE TRI-FORKS] and [QUADRUPLE TRI-FORKS] form the entire inscriptions of two ivory sticks (2795 and 2803), but a third ivory stick (2792) bears the inscription [FOUR QUOTES / QUINT-FORK].  It therefore seems safe to regard [TRIPLE CUPS] and [THREE POSTS / CUP] as alternative ways to write ‘3 CUP' (2009: 81).

Farmer et agree that this kind of sign repetition is enumerative rather than phonetic.  It may very well be.  I would point out, though, that such repetition is only quasi-enumerative in Egyptian, with three repetitions of a glyph often representing a plural form (ending in –w), alongside rare instances of quadrupling where three instances still represent the plural form with the fourth representing phonetic repetition, as in Nefer-neferu-aten, with three nfr glyphs for –neferu- and a fourth nfr representing the initial Nefer-.
Impression of a cylinder seal with Mesopotamian images but Indus inscription, MS2046 (from right):

Farmer et al further note that continuing excavations periodically turn up additional Indus symbols.  For example, there is an apparent lizard, depicted in bird’s eye view, on a seal in the Schoyen collection, a sign not included in any published list.  Another seal contains an inscription with another new symbol, a DEE with a BACK DEE overlapping it.  This type of open-endedness is to be expected in a proto-writing or non-glottographic system, especially if there is a need for new symbols to representing “owners” – just as in modern Texas, there is always room for another cattle brand to represent the owner of the livestock.
Stone block with Luwian hieroglyphs -- within the neatly incised lines separating the rows,
the hieroglyphs are grouped in a largely non-linear fashion, even though this is true writing
(from an old Turkish postcard).

An additional argument focuses on the linearity of Indus symbols.  The fact that the signs often appear in neat rows makes the script seem like writing.  But this is only a subjective judgment.  Symbols that are not linguistic have been known to appear in similar rows.  Farmer et al cite the symbols on kudurru, the foundation stones of Kassite Babylonia that bear both written texts and non-linguistic emblems of deities.  Both the signs of the text and the non-linguistic emblems are neatly aligned in rows.  To this observation I note some Navaho sand paintings, in which representations of yeis (supernatural beings) also line up in rows (Newcomb and Reichard 1975: Pl. III, with three rows of four Snake People).  In fact, the preschool child often arranges toys in a neat linear manner as a habitual form of play.  So the urge to make linear arrangements may be universal (although obviously that urge is much less strong in some people than in others).
Proto-cuneiform inscription on the Blau chisel, showing grouping of signs in compartments
rather than a strictly linear arrangement (drawn after Aruz 2003: 39).

The reverse of this argument is also of interest.  Signs of a known writing system may not necessarily be arranged in lines – whether rows or columns.  Egyptian hieroglyphs could be written either in rows or columns, from right to left or from left to right.  But even within each row or column, the glyphs do not always follow one another in a perfectly linear sequence.  Two small glyphs are frequently stacked one on top of the other in a row.  In effect, this creates a short column within a row.  And in a column, two tall and thin glyphs are frequently placed side by side, in effect forming a row within a column.  In addition, even within a word, glyphs sometimes follow a peculiar sequence.  Theophoric elements in names precede other glyphs, regardless of the sequence in the spoken language.  In King Tut’s name, for example, the last element is Amun, the name of the ram god.  But when the king’s name is written, the two glyphs spelling Amun appear in initial position, with Tut- and –ankh- following.  Luwian hieroglyphs appear even less ordered than Egyptian, in a great jumble of symbols minimally arranged in rows.  So linearity is not necessarily a sign of a writing system and writing is not necessarily linearly arranged.

Indus seal B-10 with non-linear arrangement of symbols in the inscription:
SEVEN QUOTES // TRI-FORK (or is the latter an abbreviated cult stand?).
Farmer et al also note some of the peculiarities of the apparent numerals in the Indus script.  Some of the oddest inscriptions seem to contain nothing for the “numerals” to enumerate.  With Korvink’s analysis of the various BEARER signs as terminals, how does one analyze such inscriptions as M-96 (FOUR QUOTES / POT-HATTED BEARER), or M-919 (THREE QUOTES / BEARER), or the bare "numeral" in B-10 (SEVEN QUOTES)?  Similarly, bearing in mind Korvink’s analysis of sign + BI-QUOTES as a prefix, how can one interpret M-692 (CIRCLED VEE / BI-QUOTES // FOUR QUOTES / SINGLE POST)? 
Tablet H-892A and B: note inscription on B begins with CUP on right.

Tablet H-951A and B: note the B begins with FOUR POSTS on right, ending with CUP.
The apparent numerals that do appear alongside other types of signs are sometimes described as preceding the other sign, supposedly an indication that qualifiers in general must have preceded their referents in the Harappan language.  For example, Koskenniemi and Parpola cautiously state, “Soviet scholars have expressed an opinion which we share that the Indus language appears to be fairly consistently left-branching” (1982: 12).  It is indeed common to find CUP preceded by an apparent numeral on tablets (reading right to left).  But there are also tablets with the opposite sequence, the “numeral” preceded by the CUP (H-771B right to left FOUR POSTS / CUP; H-777B CUP / TWO POSTS; H-784B TWO POSTS CUP; H-781B CUP / THREE POSTS). 

It is not just the tablets that present such contradictory data.  On seals, one frequently finds a sequence with one or another form of SEVEN preceding EF TOPPED EXIT (e.g., in H-268) (19 occurrences), but almost as common is STRIPED TRIANGLE followed by SEVEN (12 occurrences).  Similarly, some form of THREE precedes OVERLAPPING CIRCLES 18 times, but typically follows CUPPED POST or CUPPED SPOON (as in M-714).  Thus the “numerals” neither inevitably precede nor invariably follow a sign in pair-wise combinations.  If these symbols were interpreted linguistically, one would be at a loss to explain such contradictory pairings.

It is for these reasons that I accept the argument of Farmer et al that the Indus script does not represent fully developed writing.  Or, as Malcolm Hyman prefers to express it, this script is non-glottographic.  In his point of view, the important question is what relationship exists between glottographic and non-glottographic writing? (2006: 1-29).  His suggested answer is that there is no simple dividing line between these two types.  Even a society that has glottographic writing will often still continue to use non-glottographic “writing” or symbols.  As noted earlier, while we have true writing today, we also use non-glottographic numerals.  Thus, semasiography or non-glottographic systems are not failed writing, not mere historical stages on the way to true writing.  They may serve entirely different functions.  So it would be more useful to focus on the question of what functions a non-glottographic Indus system performed and how this was accomplished rather than attempting to assert that it was glottographic in the absence of clear evidence.


Aruz, J., ed. 2003. Art of the First Cities: The Third Millennium B.C. from the Mediterranean to the Indus. New Haven: Metropolitan Museum of Art and Yale University Press.

Budge, E.A. Wallis. 1920 and 1978. Egyptian Hieroglyphic Dictionary. Volume I-II. New York: Dover (1978 reprint of 1920 original, published by John Murray, London).

Dahl, Jacob L. 2005. “Animal Husbandry in Susa during the Proto-Elamite Period” in SMEA 47: 81-134.

Damerow, Peter and Robert Englund. 1989. The Proto-Elamite Texts from Tepe Yahya. Cambridge, Massachusetts: Peabody Museum of Archaeology and Ethnology, Harvard University.

Englund, Robert. 2001. “The State of Decipherment of Proto-Elamite,” available at online.

Fairservis, Walter A. 1992. The Harappan Civilization and Its Writing: A Model for the Decipherment of the Indus Script. Leiden: E.J. Brill.

Farmer, Steve, Richard Sproat, and Michael Witzel. 2004. “The Collapse of the Indus-Script Thesis: The Myth of a Literate Harappan civilization,” in Electronic Journal of Vedic Studies (11) 2: 19-57.  Available at

Hyman, Malcolm D. 2006. “Of Glyphs and Glottography,” in Language and Communication.

Koskenniemi, Kimmo and Asko Parpola.  1982. A Concordance to the Texts in the Indus Script. Helsinki: Department of Asian and African Studies, University of Helsinki.

Korvink, Michael P. 2007. The Indus Script: A Positional Statistical Approach. Gilund Press (Amazon).

Nakanishi, Akira. 1980. Writing Systems of the World: Alphabets, Syllabaries, Pictograms. Boston: Tuttle Publishing.

Newcomb, F.J. and G.A. Reichard. 1975. Sandpaintings of the Navajo Shooting Chant. New York: Dover (orig. 1937 by J.J. Augustin).

O’Connor, David. 2009. Abydos: Egypt’s First Pharaohs and the Cult of Osiris. London: Thames & Hudson.

Parpola, Asko. 2009 and 1994. Deciphering the Indus Script. Cambridge: University Press.

Potts, D.T. 1999. The Archaeology of Elam: Formation and Transformation of an Ancient Iranian State. Cambridge: University Press.

Smith, M.E. 1973. Picture Writing from Ancient Southern Mexico: Mixtec Place Signs and Maps. Norman: University of Oklahoma.

Wells, Bryan K. 2011. Epigraphic Approaches to Indus Writing. Oxford: Oxbow Books.
Old Persian cuneiform unicode document:  

1 comment:

  1. This comment has been removed by a blog administrator.