alphabetevolution

The Evolution of the Alphabet

Brian E. Colless

Writing is a sublime form of human expression and communication. As the ancient Mesopotamian scribes used to say: 'The scribal art (nam.dub.sar) is the mother of speakers, the father of scholars' (Lambert, 259; Driver, 1). Writing originally involved manipulating a stylus to make meaningful marks on clay; subsequently it also meant using a chisel to incise characters on stone or metal; ultimately it included pushing a pen or brush into an inkwell and wielding it on parchment, papyrus, or paper (and even silk, in Eastern Asia). This is characteristically a male pursuit, though the ancient Egyptians had a goddess of writing, Seshat; and the Mesopotamian ruler of the underworld, Queen Ereshkigal, had a female scribe as her right hand person.  

In seeking to chart the course of the evolution of the alphabet, we must first address the fundamental question: Was the pen created for commercial or religious purposes? Or even for legal or literary ends?  

Origins of Writing

Apparently writing was invented five thousand years ago in ancient Sumer, in the towns of southern Mesopotamia, between the Tigris and Euphrates rivers, near the Persian Gulf, that is, in ancient Iraq (Walker, 7-13). Uruk and Ur were two of its oldest cities, and both are known in the Bible as seats of civilization (Genesis 10:10,11:31). Later came Babel (Babylon to the Greeks), with its great tower (Genesis 11:1-9), a focal point for astrological astronomy and numerological mathematics. (Our 60 seconds and minutes, and our 360 degrees, are legacies of Sumeria and Babylonia; their mathematical system had 60 as a basic unit; and they had lunar months and years. The Egyptians gave us the solar calendar, the year of 365 days.)  There are Chinese claims of even older examples of writing, but the Chinese systems would not have  influenced the development of the alphabet, which is the subject under discussion here.

The oldest writing from Sumeria was inscribed on clay with a wooden stylus (Walker, 7, 22-26). A Mesopotamian tradition says that writing on tablets was introduced by King Enmerkar, whose messenger was ‘heavy of mouth’ (Schmandt-Besserat, 1-2). Diplomatic correspondence was certainly one of the uses to which writing was put, but eighty-five percent of the inscribed clay tablets from the early levels of Uruk are economic documents (the remainder being lexical lists; Walker, 11). These recorded incoming and outgoing, of food, livestock, and textiles. They are thus accounts; that is, they belong in the field of accountancy. This seems to be the answer to our question: this evidence points to a commercial origin for writing. Yet these documents are in a sense religious, in that they are mostly temple records. 

Here we must note that before writing was invented tokens (made of clay or stone) were widely used, for counting sheep, and oxen, and various commodities; these tokens were the prototypes of some of the symbols used in accountancy and writing, and they were current many thousands of years before the earliest writing, as demonstrated by Denise Schmandt-Besserat (1992). The tokens were eventually impressed on clay, rather than being collected in a container; and then signs for numbers were invented. This produced a system of notation, which had symbols for numbers, quantities, and commodities; but this was not true writing. The evolutionary principle thus seems to be that reckoning comes before writing. (The modern analogy would be the computer, which basically works with numbers, but also processes and stores pictures and words.) Note, however, that there are no examples of economic accounts from the Maya civilization of Mesoamerica; the Maya syllabary was for calendrical, historical records (Houston 1989, 22, 48). The alphabet was adopted in Greece, around 800 BCE; and the Greeks apparently did not have commercial purposes in mind either (Senner 1989, 9). The first major collection of alphabetic Greek inscriptions are what Barry Powell has called the erastic graffiti from Thera. But Powell suggests that the Greek alphabet was devised specifically for putting Homer's Iliad and Odyssey into writing (Powell 1993). Nevertheless the oldest examples of syllabic Greek manuscripts, namely the Linear B tablets of Greece and Crete, were administrative accounts and records, not literary or historical texts (Chadwick 1987).

The ancient West Asian tokens, even when they became impressions on clay, were only precursors of writing. The definition to be followed here is this: writing is visible speech (De Francis 1989; Diakonoff, 6). Writing is a technique or system for reducing speech sounds to visible characters; a script is a set of characters (graphs, graphemes) for representing coherent speech. An alphabet is a minimal set of signs for representing the sounds of a language in script. 

The Sumerian script (which was complex and therefore not an alphabet) was devised for the Sumerian language, though this is not absolutely certain, since the people in the Elam region of ancient Iran were also using it from the beginning (Walker, 7-9). Subsequently it was adopted and adapted by the Akkadians, a Semitic people who moved into the fertile region of the two rivers (Walker, 15-18). While the Sumerian writing system was not created for legal or literary ends, it was nevertheless exploited for recording law-codes (like Hammurabi's, which was engraved on a diorite/greenstone pillar), and also for inscribing contracts on tablets. Such documents were not only legal but also religious. Thus, Hammurabi names the gods and their temples, before getting down to promulgating his legislation; and in treaties, the gods were invoked as witnesses, and as guardians of the oaths.  

Literature came into existence when Sumerian and Akkadian hymns, myths, and legends were written down; two notable Akkadian examples are: the Babylonian myth of creation (Enuma elish), in which the deity Marduk is exalted as king of the gods and creator of the world; the epic of Gilgamesh, the story of a king of Uruk, who goes in quest of fame and immortality. Whether the main protagonists are human or divine, the gods are actively involved at every step of these stories. A significant Sumerian myth relates the transfer of the arts of civilization from Eridu to Uruk: Enki, the culture-deity of Eridu, becomes intoxicated with wine, beer, and the beauty of Inanna (Akkadian Ishtar), the Mesopotamian Venus, and he magnanimously hands over all the components of civilization to the goddess, who quickly takes them to her people in Uruk. Included in this treasure trove was ‘scribeship’, the art of the scrivener.

The celebrated Egyptian hieroglyphs sprang up in the Nile Valley, at about the same time as writing appeared in Mesopotamia, 3000 years before the current era (BCE), and possibly in imitation of the script used in ancient Iraq and Iran (Davies, 10, 37-40). Both writing systems, Mesopotamian cuneiform and Egyptian hieroglyphic, began as pictographs/pictograms: pictures standing for the things they portrayed. But the pictographs soon became stylized, so that the objects they represented were no longer recognizable. The Mesopotamian pictographs disintegrated into clusters of tadpoles, or wedge-shaped marks inscribed in clay; so we call them cuneiform characters, wedge-shaped signs (Latin cuneus, a wedge). The Egyptian hieroglyphs were reduced to cursive strokes of the pen or brush (like Chinese characters, but even more abbreviated): first the HIERATIC script, and later the DEMOTIC script, both written on papyrus. However, the Egyptian hieroglyphs continued to be engraved (on stone and metal) in their picture forms, right to the end (Davies, 10-24).

Evolution of Writing

If we are invoking the idea of evolution, then we should establish some evolutionary principles, or analogies from the natural world (Dennett, 175-6). First, nature is pragmatic and does not care about origins; if something works then so be it. And in the ancient Mesopotamian and Egyptian scripts, as also in the early alphabet, the characters functioned satisfactorily even when their original pictorial features were lost.  

Secondly, multifunctionalism is a possible phenomenon in nature: particular elements in a single organism may have more than one function. In contrast, systems designed by humans usually seek to avoid side-effects; but in writing systems we certainly find multiple functions for single elements. At all stages of the evolution of the alphabet a character may be multifunctional. Thus the letters of the Greek alphabet were also used as numerals (in a system that was more complicated than our use of a, b, c, d et cetera to itemize points in a series of propositions). And when the letters of the alphabet were still recognizable pictures they could also stand for the thing they pictured, thus representing a whole word, not merely a single sound. They could each act as a logogram (though not as an ideogram with a variety of references). That is how I propose to define and differentiate logograms and ideograms: logograms normally represent only one word each, while ideograms denote a number of related concepts.  

In Egypt and Mesopotamia the original pictographs were used as logograms and ideograms (or logographs and ideographs) representing words and ideas (Davies, 30-37; Walker, 10-11). As a logogram, a picture of a foot could mean ‘foot’ (a noun) in any language, we would assume; and as an ideogram it would denote actions of feet (verbs). However, there are complications and seeming anomalies, caused by the human mind's propensity to free association of ideas. As an ideogram the Sumerian foot sign covered such pedal activities as ‘stand’, ‘walk’, ‘go’, ‘come’, and ‘bring’ (Walker, 10; Labat, 117, No 206), but not ‘foot’.   Another example of an ideogram is a set of three lines, for the idea of triple, and for the numeral 3 (‘three’, like the Roman III). A common English usage (in advertising and in graffiti) is the heart sign: it says 'heart' on a playing card, and the verb ‘love’ in ‘I HEART (=love) U, and’Jesus HEARTs you’ (where the s is a phonetic complement.

Pictographs could also function as determinative signs, semantic markers: for example, an Egyptian hieroglyph (D54) shows two legs walking, and when placed after another word it indicates that this is a verb of motion (or, paradoxically, lack of motion, such as 'stop' or 'linger').

Then came the REBUS principle: the reader merely focusses on the sound of the word or idea depicted in the sign, and accepts it as another word (a homophone) or simply as a syllable in another word (a syllabogram, in effect).  

Here are some examples, using English analogies. A picture of a one cent coin has countless uses when we put the following English statement into rebus writing (we will need ten mental cents for the task, and one of them will in fact be a simple logogram, not a complex syllabogram). 

    'A sententious sentence: Being sentimental, though having no sense of decency, I gave the poor innocent child one cent, and sent him off after the recent scent of hot pies'. 

Now draw the following rebus pictographs on your mental slate, and read out the message: the numeral 2, a honey bee, an oar (from a rowing boat), a knot in a string, another 2, another bee. 'To be or not to be', that is the answer.  And that is the rebus principle at work. The result is called logo-syllabic writing: the signs stand for whole words or syllables. If we borrowed the Sumerian script for English (the Semitic Babylonians and Assyrians did, as also the Indo-European Hittites of Asia Minor) we could write the greeting ‘Good day! ‘.

    With the signs GU (‘ox’, an ox-head) and UD (picture of the sun rising, representing ‘sun’, Utu the sun god, or ‘day’) we could have the two syllabograms GU and UD producing ‘good’, and  UD also as an ideogram for ‘day’ (rather than ‘sun’):

    GU-UD UD (=‘day’).  ‘Good day.’

But the word for 'ox' is actually GUD, so GUD UD would be sufficient.

Or we could simply use a logogram (a Sumerogram) for ‘good’ (DUG):

    DUG (=‘good’) UD (=‘day’).   ‘Good day.’

One last example of Sumerian ingenuity will be noted: the Sumerian pictograph for 'head' was modified by the addition of lines around the mouth region; as an ideogram it then represented KA 'mouth', ZU 'tooth', the verb DU 'speak', and the related nouns GU 'voice' and INIM 'word'. All these words were then potentially available as phonetic syllabograms, for writing other words, but in practice the sign's primary value was ka (M.W. Green in Senner, 46-47; Labat, 49-50, No 15) 

The multiple functioning of particular signs is a feature of logo-syllabic writing, whereby the graphemes denote either words (logograms, ideograms) or syllables (syllabograms). The human brain is certainly capable of processing such complications (Reiner 1973), but we would expect that there would have been an impetus to devise a simpler system (as there is in modern China and Japan). Of course, Mesopotamian and Egyptian scribes who had laboriously mastered the technique, and whose livelihood depended on its very complexity, would resist change. Nevertheless, simplification occurred in centres beyond the geographical spheres of their influence.

Syllabic Writing

 A surprising and fundamental difference existed between the Mesopotamian cuneiform script and the Egyptian hieroglyphics: the Egyptians took no account of the vowels in their writing. To explain this by analogy, using English words: if we used a simple picture of the city Paris to represent the name Paris, we could also use it (in accordance with Egyptian principles) for the word porous, and as part of the word parasite, because only the consonants count. The vowels are ignored. But in Mesopotamian cuneiform writing we would have to break the word porous into syllables, and use a separate sign for each syllable (though there is no sign for po, so the first syllable has to be pu ): pu-ru-us or pu-ru-su (with the last vowel unsounded, ignored). 

Around 2300 BCE, the scribes of ancient Ebla (in Syria) were writing their own North Semitic language in Sumero-Akkadian cuneiform script; they mostly used Sumerograms (Sumerian logograms), which were read as Eblaic words (thereby making it hard for us to determine what their own Eblaic words were); but when they spelled out a word they tended to avoid closed syllables (ending in a consonant, such as us in pu-ru-us for porous) and employed open syllables (consonant plus vowel, as in pu-ru-su for porous), with some syllables having dead vowels, which are ignored and left unsounded. For example, assammu (a drinking-vessel), commonly written as-sa-am-mu in Mesopotamia, becomes a-sa-mu-mu in Eblaic 'open-syllable orthography' (Saggs, 75; Gordon, 129). Notice that the first mu has the same vowel as an adjacent (following) syllable, but its vowel is silent. In Eblaic ni-ka-ra-du, for Mesopotamian nin-kar-du, the syllable with the unsounded vowel is ra, which silently echoes the preceding syllable ka.

This idea of muted vowels in syllabic spelling marks a new phase in the development of writing. It was a feature of the Linear B script of Crete and Greece in the second millennium BCE, where tri  would be written ti-ri (Chadwick, 23, 26). Compare also synharmony of the dead vowel with the preceding vowel in Maya syllabic spelling (Houston, 39). However, both the Cretan and the Mayan systems were preceded by another Semitic script, namely the West Semitic logo-syllabary, which also appeared around 2300 BCE.   (The Cretan logo-syllabary probably used it as a model, and it is not beyond the bounds of possibility, given the propensity of the Phoenicians for undertaking long voyages, that American writing systems were inspired by their invention.)

Open-syllable orthography can be viewed as a compromise between the Egyptian system, with absolutely no vowels represented in its hundreds of signs, and the Mesopotamian system of hundreds of syllabograms (when a hundred or less would suffice, as in Linear B). And so, probably in the Bronze Age city of Gubla (now Jibeil in Lebanon, and known to the ancient Greeks as Byblos) a new script was invented for the language of the West Semites (that is Canaanian, or Phoenician, or Hebrew, as it is variously known). Examples of this script have been found in Syria-Palestine, Sinai, Egypt, and Italy (Colless 1998). Gubla was on the Mediterranean coast, a Phoenician port, having very ancient links with Egypt and Babylonia. To construct the new Gublaic syllabary, Egyptian hieroglyphs were borrowed (as far as possible, but it was not always feasible). However, the Gublaic script followed the Mesopotamian practice of indicating vowels in its syllable signs. Its inventory was limited to less than seventy, in stark contrast to the Egyptian and Babylonian systems. This economy was achieved by having only open syllable signs (consonant plus vowel), and only three basic vowels (a, i, u): ba bi bu, sa si su, and so on (but not bat bit but, or at it ut). The idea of dead vowels in some syllables was part of the scheme: bint would be written bi-ni-tu or bi-nu-tu, with two unsounded vowels. (For examples of Gublaic syllabograms, see the first sign in each row of the Phoenicia column, "BS", Byblos syllabary,  on the table of signs.)  

To create the Byblos syllabic signs the principle of ACROPHONY was followed ('the top sound' principle'; Greek akron means 'top, highest point', in this case the first syllable of the word that goes with the sign; cp. acronym). Thus the Egyptian hieroglyph for an oil jar was employed as the syllable du, because the West Semitic word for it was dudu. The word for 'door' was daltu (the Greek letter Delta got its name from this), so the picture of a door stood for its initial syllable da.  Because gamlu was their word for 'boomerang' (sceptics need to be told that Tutankhamun had boomerangs in his tomb) the "throwstick" hieroglyph represented ga (Colless 1992, 63-66). 

It is not known precisely when the West Semitic syllabary was invented (attested at Umm el-Marra [Tuba?] around 2300 BCE), but it presumably came into existence before the alphabet, and it was another step in the evolution of writing, leading up to the alphabet. It is widely held that the Byblos script has not been deciphered yet. Nevertheless, George Mendenhall of Michigan claims to have succeeded, after spending thirty-seven years trying to decode the small collection of documents on copper plates and stones (papyrus texts would presumably have existed and perished long ago). I have taken Mendenhall's results (published in 1985) and I have done my own tests on them (my modified scheme was published in Abr-Nahrain in 1992). Even if I am wrong, partly or entirely, we still have the Cretan scripts as analogies for what a Bronze Age syllabary was. And when I compare the Byblos syllabary of about seventy signs with a similar West Semitic script which has only two dozen hieroglyphic pictographs, I feel fairly confident that I am observing a series of mutations, of vital significance for the evolution of the alphabet. This more compact hieroglyphic script was the proto-alphabet, the West Semitic  pictographic alphabet (Colless 1992, 96-98). 

Alphabetic Writing 

The oldest collection of proto-alphabetic texts comes from the Sinai Peninsula, in and around the copper and turquoise mines that were anciently exploited by the Kings of Egypt (Colless 1990; Sass 1988, 8-50). Over the centuries the Egyptian expeditions left hundreds of hieroglyphic inscriptions. And at some point in the second millennium before the current era, the proto-alphabetic inscriptions were written: chiselled onto sandstone, or inscribed on the interior and exterior walls of the mines. The likeliest assumption is that they were the legacy of West Semitic workers, possibly prisoners-of-war (Colless 1990, 18-19), slave-labourers, like the Hebrews who were slaving over hot stoves (brick-kilns) for the Egyptians. This analogy and the word 'kiln' (Hebrew kibshan 'furnace') are important for my decipherment: the Semites functioned as metalworkers, as shown by the remains of their metallurgical equipment and the mention of melt-furnaces in their inscriptions (Colless 1990, 6)  

Bereft of female company, the workers in the mines, and their overseers, had to practice sublimation on their mental and metal images of goddesses, either the mother goddess, or the goddess of love, according to their needs. The Egyptians saw themselves as loved by Hat-hor, the goddess with beautiful cow ears. When I was a student of ancient and modern languages at Sydney University, I often gazed at a huge block of stone with Hathor's alluring features carved into it: fascinating ears, seductive smile. In the nearby Nicholson Museum was a facsimile of the Rosetta Stone, which had provided Champollion with the key to the decipherment of Egyptian scripts, with its triple text in hieroglyphic and demotic Egyptian, and also in Greek, written in the Greek alphabet. The names on the Rosetta Stone led to its decipherment, and that is also how the proto-alphabetic code was cracked. The decipherment was initiated by Egyptologists, and it was Hat-hor who inspired them.  

Ironically, it was an alphabetic (Greek) text, on the Rosetta Stone, which gave the solution to the hieroglyphic mystery, and it was a hieroglyphic text that unlocked the door to the proto-alphabet. We now come to the riddle of the Hathor sphinx. William Matthew Flinders Petrie found the Sinai inscriptions in 1905 (Petrie 1906). He dated them to the time of Queen Hatshepsut, 15th century BCE. In the temple of Hathor, near the mines, a small, red sandstone sphinx was brought to light. It had characters inscribed on it, all pictographic, but only one cluster of them could be read as Egyptian hieroglyphs, representing a phrase that occurred repeatedly in the Egyptian inscriptions in the area: 'beloved of Hathor, the turquoise lady'. Flinders Petrie had his own fixed views on the origin of the alphabet, and he did not believe it was derived from Egyptian hieroglyphs. 

Nevertheless, two other Egyptologists pursued this line, namely Kurt Sethe and Alan Gardiner. In 1915 Gardiner looked at the characters on the sphinx, saw the head of an ox, and thought it must be an ’aleph, the first letter of the Hebrew and Phoenician alphabet. ’Aleph means 'ox', and it is the word behind the Greek letter-name Alpha. And the letter Alpha, as also the Roman A, which we use as the first letter of the alphabet, is clearly an ox-head (turn a capital A upside down and gaze at the horns and snout of a bull or a cow). Here is a Hathor-cow-connection, but more significantly Gardiner knew that in the Semitic city Byblos, Egyptian Hathor was identified with Semitic Ba‘alat (Greek Baaltis). The -at ending of Ba‘alat shows that she is the female counterpart (and consort) of the male weather-god Ba‘al, namely Astarte or ‘Athtart. Ba‘al means 'Lord, Master', Ba‘alat means 'Lady, Mistress'. And Gardiner noted a particular group of four letters recurring in the inscriptions. 

The first character was the ground plan of a house (sometimes with a doorway, or even a porch). That had to be beth ('house'), the second letter of the Hebrew and Phoenician alphabet, and the origin of Greek Beta. The second sign was a human eye (‘ayin), a guttural consonant, but it became the vowel o in the Greco-Roman alphabet. Third came a letter that has not changed much in its long history, a shepherd's crook (lamed, Greek Lambda); its bent end is still there in cursive L (l). The fourth letter is a cross, T, the last letter of the West Semitic alphabet. Hence B‘LT (Ba‘alat). Clearly, there were no vowel signs  in the early alphabet. A few years later Eisler recognized that the sequence of signs preceding the name Ba‘alat was m’hb (mu’hab), meaning 'beloved'; 'beloved of Ba‘alat' corresponded to 'beloved of Hathor' in the Egyptian inscriptions (Sass, 12-13).

In the proto-alphabetic texts the Egyptian model was followed:  concentrate on the consonants, and disregard the vowels (Millard, 394). The principle of acrophony was applied, as with the Byblos syllabary, but instead of taking the entire first syllable of the name of the character (with its vowel intact) only the initial consonant was sounded. The 'dead vowel' principle, we might say, was thus carried to extremes: all vowels were unsounded and unrepresented in proto-alphabetic writing. This meant that only the person who wrote a particular inscription really knew its meaning for certain, such was the ambiguity of this shorthand system. 

How had Gardiner achieved his breakthrough? By 'sequencing', a method of analysis later to be practised by geneticists: the researcher looks along the chain of 'letters' (whether DNA or ABC) and picks out recognizable 'sequences'. And, as in DNA decoding, variations can be found which nevertheless add up to the same thing. Thus 'beloved of Ba‘alat' occurs in such combinations as m’hbb‘lt (the fullest form), mhb‘lt (the most abbreviated), and m’hb‘lt (Colless1990, 14-15).

Another point for comparison is the standard three-letter codon in genetics (a code unit that determines amino-acid sequence). In Semitic languages (such as Hebrew and Arabic) words are built on tri-consonantal roots. Thus Ba‘al and Ba‘alat are constructed from the three-letter root b‘l (meaning 'possess'), and mu’hab ('beloved') is from the root ’hb ('love'). The interpreter of Semitic texts searches along the lines of characters for the three-letter roots, but a trap is that some words have only two consonants.  

There are other problems in sequencing the Sinai inscriptions. There is no set direction for the writing to go: N, S, E, W are all possible. Worse, there is no punctuation: the writing runs on without dots, bars, or gaps to separate sentences, phrases, or words. And of course there are no signs for vowels. Semites, whether ancient Phoenicians and Israelites, or modern Arabs and Israelis, are quite accustomed to reading unvocalized texts (but dots and dashes are added to indicate vowels in the Hebrew Bible and the Arabic Qur’an).  

In 1915, Alan Gardiner reported his discovery at a meeting of the British Association for the Advancement of Science, and this was published in the Journal of Egyptian Archaeology in 1916, under the title 'The Egyptian origin of the Semitic alphabet'. In 1929, Gardiner delivered another lecture on the subject, in which he told how had made his breakthrough; but he also expressed his concern over what other people were doing with the Sinai texts. He quoted a few strange-sounding interpretations on a statuette dedicated to Ba‘alat, and declared their attempts at translation unconvincing; as an outside spectator he thought that 'very little real advance' had been made in interpreting the Sinai inscriptions. In short, he declared that 'the Sinaitic material is too slight, too badly preserved, and too ambiguous to admit of any very serious attempts at complete interpretation'.

While he yet spake there came good tidings of great joy: a new three-letter pictographic inscription was found, in 1929, in Gezer, an important city of ancient Canaan (Colless 1991, 31-32; Sass 1988, 55-56). This showed that the proto-alphabet was not just a phenomenon of Sinai. The likelihood was, in my view, that the proto-alphabet had originated in the cities of Canaan and had been taken to Sinai by the prisoners-of-war who worked as slave-labourers in the Egyptian mines.  

More recently, the proto-alphabet has appeared in the Wadi el-H.ôl, in the Western Desert of Egypt, near Thebes (Luxor), two one-line inscriptions (vertical and horizontal) apparently dating from the Middle Kingdom period (about 2050 - 1550 BCE), and this has seriously been proposed as the place where the proto-alphabet was invented, not Sinai or Canaan (Darnell 2005). Certainly, there are other examples of  proto-alphabetic writing  from elsewhere in Egypt, notably a set of ostraca collected by Petrie in Luxor and published in 1912, but they have been ignored until now; one of them bears a copy of the original letters of the proto-alphabet.