Empires of the Plain: Henry Rawlinson and the Lost Languages of Babylon

The very first writing evolved in Mesopotamia from the need of accountants and bureaucrats to keep a visual check of goods entering and leaving temples and palaces. Small clay tokens dating from 8000 BC appear to have been an early tally system and a precursor of writing. They have various geometric shapes, such as spheres, cones and discs, perhaps representing different commodities. In the mid-fourth millennium BC, tokens were sometimes placed inside hollow clay balls or envelopes and were sealed with cylinder seals.

Used in Mesopotamia for over three thousand years, cylinder seals were invented around 3600 BC as an aid to bureaucracy in the vast city of Uruk in southern Mesopotamia. These seals are small cylinders, usually made of imported stone and carved with intricate designs, especially scenes of everyday life. When rolled across damp clay, they left a continuous impression, and both seals and clay sealings have been found. As well as on clay balls, seals were used on clay tablets and on lumps of clay attached to cords securing door-bolts, bags, sacks, boxes, jars and other containers, as a deterrent against theft. The sealed clay balls may have accompanied deliveries of merchandise (acting as bills of lading), whose contents could be checked by breaking open the balls to reveal the tokens. Some balls have marks on the outside that seem to indicate the number and type of tokens they contained, but this information could also be recorded on flat clay tablets, and the earliest ones – termed numerical tablets – date from 3500 BC and had impressions of tokens and cylinder seals similar to those on the clay balls that they replaced.

The most primitive form of recognizable writing was a book-keeping system done on clay tablets, with simple signs for numbers and pictorial signs (pictographs) to represent what was being counted or listed, such as oxen or barley. At this stage, the wedge-shaped stylus producing distinctive ‘cuneiform’ writing had not come into use. Instead, signs were written with a stylus that had a circular end to make impressions representing numbers and a pointed end to draw linear pictorial signs, and this writing is termed ‘proto-cuneiform’. Tablets with a proto-cuneiform script date to 3300–2900 BC. Since signs were written on damp clay, scribes could only produce stylized sketches rather than the realistic images used in Egyptian hieroglyphs. Two wavy lines, for example, represented water, while the outline of a head of an ox represented an ox. Pictorial signs were also used as ideograms, to represent an associated idea. For example, a picture of a mouth might also mean ‘to speak’. About 1,200 signs are known, but many are not understood today. On the clay tablets, groups of proto-cuneiform signs were written relatively randomly within square or rectangular boxes. The boxes themselves were arranged in horizontal rows that were read from top to bottom and in a right to left direction.

The amount of information that could be expressed by this sort of writing was severely limited – most proto-cuneiform tablets were concerned with book-keeping, although about 15 per cent of the surviving tablets are lists of words, such as the names of animals and cities, and were probably used in the training of scribes. As each pictograph or ideogram represented an entire word that could be understood universally, like those used today at airports, the language spoken by the people who made the pictographs is uncertain, but was probably Sumerian. In south-west Iran, once known as Elam, a script composed of numerical and other signs has been found on similarly early clay tablets dating to 3100–2700 BC. It seems to be more developed than simple picture signs, but remains poorly understood. It was originally labelled proto-Elamite on the assumption that it was an early form of the Elamite language that was written down a few hundred years later.

The next development was to write words using the sounds of syllables, a more sophisticated method that enables specific languages to be identified. The earliest language to have been written on clay tablets and the earliest known language in the world is Sumerian, dating from at least 2800 BC – or several centuries earlier if proto-cuneiform is accepted as Sumerian, and so earlier than Egyptian. Sumer was the southern part of Babylonia in the period 3000–2300 BC, and stretched from Nippur (south of Babylon) to the Persian Gulf. The Sumerians called their territory kiengir (‘homeland’), but later their Akkadian neighbours called it Shumeru, from which the modern name Sumer is derived. Sumerian is not related to any known family of languages, and it was possibly the only one of its family to have been written down, with the others dying out before writing was invented. Nothing is known about languages before Sumerian.

By 2600 BC major changes had taken place with the signs used to write Sumerian. They were now written as if turned 90 degrees to the left, so that the outline of a head of an ox was turned on its side and the two horizontal wavy lines representing water became vertical lines. They were also now written from left to right, not right to left. The other major change was that signs were no longer incised in the clay, but were impressed using a reed stylus with an angular end, forming the distinctive cuneiform (‘wedge-shaped’) signs. This was a more rapid way of writing, with the stylus being pushed into the damp clay rather than used to incise lines. Because of this change, signs became much more stylized, so that the sign for an ox was composed of a few impressed lines, barely resembling the original abstract outline of the head of the ox.

These cuneiform signs could still be used as pictographs, so that the stylized sign for a mouth meant ‘mouth’ (ka in Sumerian), and they could still be used as ideograms, so that the sign for a mouth also meant ‘tooth’ or ‘word’ (zú and inim in Sumerian). Many Sumerian words had only one syllable, such as ud, ‘day’, and sometimes even a single vowel, such as a, ‘water’. The signs for these words began to be used phonetically to spell out syllables of different words, regardless of the original meaning of the sign, such as the sign for barley, pronounced she, being used where the syllable she was required. In English an example would be a picture of a ring used to spell the ‘ring’ sound values of ‘bring’, ‘ringleader’ and ‘daring’. As words could now be spelled with syllables, there was no longer a need to have a separate sign for every Sumerian word, and so the number of signs dropped to around six hundred. This was still a very complicated system when compared with modern alphabets of around twenty-five or twenty-six letters, and so knowledge of writing was restricted to specially trained scribes, with the rest of the population remaining illiterate.

On the early Sumerian tablets, the cuneiform signs were grouped randomly in boxes, which were arranged in horizontal rows that were read from the top of the tablet to the bottom, but each box was now read from left to right, not right to left. With the increased use of signs for syllables, the written language became more structured, and grammatical elements developed. More complex words could be expressed, and because word order became important, signs began to be written in a single horizontal line, from left to right. Even so, there was no punctuation, nor any spaces between words.

Because Sumerian cuneiform signs started off as pictographs that were subsequently used as ideograms and syllables as well, almost every sign acquired several different functions. Many signs (termed polyphones) had several alternative sounds. For example, the sign

is du, meaning ‘leg’, but the same sign can have other associated meanings with different pronunciations, such as gub, ‘to stand’, gin, ‘to go’, and túm, ‘to bring’. To get around this problem, the correct reading could be emphasized by adding another cuneiform sign, called a phonetic complement, comprising the final consonant and a vowel (usually a). This sign was not pronounced, but indicated what word was meant. For example, when this particular sign was to be read as gin, a sign for na was added, which cuneiform scholars write as gin(na) or gin

.

Several Sumerian signs were pronounced the same way (like flour and flower in English). These signs are termed homophones – having the same sound. For example, there were ten different signs for the word or syllable pronounced tum. In modern transcriptions the particular word meant is shown by the addition of accents and numbers (diacritic signs). For example, several different signs were pronounced gu. The one meaning ‘ox’ is written in transcriptions as gu

. Scribes also added signs called determinatives (or classifying signs) before or sometimes after a word to indicate the category to which a word belonged, so that its meaning was clarified. The sign ki, for example, indicated that the adjacent word was a place-name and dingir the name of a god. These determinative signs were not pronounced, but were present merely to show the meaning of the words.

Sumerian ceased to be an everyday spoken language by about 2000 BC, but scribes continued to copy out texts and word lists, often with Akkadian translations, because Sumerian became a prestigious and scholarly dead language, like Latin in the Middle Ages. Akkadian, the oldest known Semitic language, belonging to the same family of languages as Hebrew and Arabic, had become the everyday spoken language. The term Semitic was coined in the eighteenth century because the speakers of these languages were believed to be descendants of Noah’s son Shem or Sem. Originally used alongside Sumerian, Akkadian was first written down from around 2500 BC. Although a Semitic language, the cuneiform writing system for Akkadian was based on Sumerian, despite the two languages being vastly different. Early cuneiform decipherment did not tackle Sumerian, as its existence was not initially recognized. Akkadian (‘the tongue of Akkad’, lishanum akkaditum) takes its name from Akkad (or Agade), which was founded as the capital city of the new empire of Akkad around 2300 BC by King Sargon, after he united several independent city-states in northern Babylonia and Sumer. The city of Akkad has not yet been discovered, but it probably lay north of Babylon.

Most Akkadian words had more than one syllable, and the cuneiform signs used to spell out words phonetically were either single vowels such as a, consonant-vowels such as tu, vowel-consonants such as an or consonant-vowel-consonants such as nim – never single consonants. Sumerian signs were frequently adopted as syllables or to represent entire Akkadian words. For example, the Sumerian sign

an, meant ‘sky’ or ‘heaven’, and this same sign was adopted for Akkadian, but in that language was pronounced as shamu. The same Sumerian sign could mean a god, dingir, which was also adopted in Akkadian, but pronounced ilu. Signs taken from Sumerian are now called Sumerograms. Cuneiform scholars today write Sumerian words in lower-case Roman script, Akkadian words borrowed from Sumerian in UPPER-CASE Roman script and Akkadian words in italics in an attempt to lessen the confusion.

As in Sumerian, a few Akkadian signs were used as determinatives and placed before or after words to clarify the type of word (such as a place, woman, god), and these signs were not pronounced. Phonetic complements functioned in a similar way to those of Sumerian cuneiform, but were not so widely used.

By 2000 BC about six hundred Akkadian signs were used, but most signs had two or more values or readings, representing a syllable, an entire word or a determinative. Some signs (the polyphones) had more than one phonetic value or syllable, such as the sign

, which can represent the syllables ur, lig or tash, and several different signs (the homophones) shared the same sound, such as

which all represent the sound ur. As with Sumerian, scholars today show a sign’s value by a system of accents and numbers: the most common homophone in a group has no notation, the second an acute accent over the vowel, the third a grave accent, and the fourth and following have numbers, as in ur, úr, ùr, ur

and ur

, called ur-one, ur-two, ur-three, ur-four, ur-five and so on. They are all pronounced in the same way.

Sumerian had a great influence on the written form of Akkadian, such as the verb occurring at the end of the sentence, which does not happen in other Semitic languages. However, verbs in Akkadian were not constructed like those of Sumerian (which had a fixed root word to which prefixes or suffixes were added). Instead, they had a root of three consonants (triliterals), which changed internally according to the meaning, mainly with the addition of different vowels. This is similar to English: the verb ‘to write’ can have various forms, such as written, writes and wrote – but w, r and t remain constant. Many Akkadian nouns ended in ‘m’, such as sharrum (king), but this ending was dropped towards the end of Old Assyrian and Old Babylonian, so that the word became sharru. There were no spaces between words, but there was occasional punctuation, such as an upright wedge

to indicate the beginning of a sentence. The writing was read from left to right, and larger clay tablets could be divided into columns, like a modern newspaper, which were also read from left to right. On the reverse of tablets, though, the order of the columns could be left to right or right to left. Horizontal lines often separated each line of cuneiform writing.

There were three main Akkadian dialects, known today as Old Akkadian, Babylonian and Assyrian, and all used slightly different cuneiform scripts. In reality they were so similar that the terms tend to be interchangeable, and today they are studied as a single language. As with any other language though, Akkadian changed over the centuries. The Old Akkadian dialect dates to 2500–2000 BC, and under King Sargon it replaced Sumerian as the official language of administration. Only a century after its foundation, the Akkadian Empire collapsed, and for the next few centuries southern Mesopotamia experienced incursions from neighbouring tribes and was ruled by dynasties from cities such as Ur and Babylon, even though the kings still described themselves as rulers of the lands of Sumer and Akkad.

From 2000 BC it is possible to distinguish between the dialects of southern Mesopotamia (Babylonia) and northern Mesopotamia (Assyria). Babylonia incorporated what was formerly Sumer and stretched from the Persian Gulf northwards to the present city of Baghdad. Babylonian, the dialect of this region, is usually subdivided into Old Babylonian (2000–1600 BC), Middle Babylonian (1600–1000 BC), Neo-Babylonian (1000–600 BC) and Late Babylonian (600 BC to AD 75), while the term Standard Babylonian is used for the version of Old Babylonian that was preserved after 1500 BC by scribes in Babylonia and Assyria. From 1400 BC cuneiform, especially Babylonian, became the international language, the lingua franca, of diplomatic relations and trade over a vast area from Asia Minor to Egypt. Literacy rates within the population were still low, as the written language remained difficult.

The Assyrian dialect of the Akkadian language was contemporary with Babylonian, but was spoken in northern Mesopotamia. This region, known as Assyria (after the town of Ashur or Assur), stretched from what is now Baghdad northwards to the Anatolian mountains. Its main towns were Ashur, Nineveh, Nimrud, Khorsabad and Arbela, and at first Assyria was a collection of independent city-states. It became a powerful military state, expanding its territories and even invading Babylonia and sacking Babylon in 1235 BC. After 1100 BC Assyria went into decline, but from 930 BC the Neo-Assyrian Empire emerged as the dominant force in the region, conquering and annexing territory as far as Israel, Judah and Egypt. Many cuneiform inscriptions from this period have been found in vast library archives of clay tablets. In 612 BC the empire collapsed when Nineveh fell to the Babylonians and Medes.

The Assyrian dialect is usually subdivided into Old Assyrian (2000–1500 BC), Middle Assyrian (1500–1000 BC) and Neo-Assyrian (1000–600 BC). From the eighth century BC, Aramaic – the Semitic language of the Aramaeans, a nomadic tribe from the Syrian desert – became widespread as a spoken language, gradually replacing languages such as Akkadian. Scribes of cuneiform and Aramaic are depicted in sculptured reliefs working side by side at this time. The Aramaic writing system, based on the Phoenician alphabet, was much more simple and could be written with pen and ink on materials such as parchment and papyrus. It soon began to be adopted in place of cuneiform, and Aramaic became the international language of diplomacy and administration, while Akkadian became a literary and scholarly language.

From the sixth century BC Persia (modern-day Iran) began to expand its already immense empire westwards, first into areas like Elam and Babylonia where cuneiform was used and later as far as Egypt and Greece. Elamite, a non-Semitic language not closely related to any other, is first seen around 2300 BC and became an official language of the Persian Empire. It is known mainly from hundreds of clay tablets found at Susa, the city that became the summer capital of Darius the Great, and also at his new capital Persepolis, as well as on monumental inscriptions such as at Bisitun. Not content with adopting Babylonian and Elamite cuneiform, Darius also invented a system of cuneiform for writing down his own language of Old Persian, which had never before been written down. This was the first time in antiquity that a complete writing system had been invented, rather than gradually evolved. Old Persian cuneiform began to be used from early 520 BC in the inscription at Bisitun, and Darius and his successor Xerxes had many of their achievements recorded in other trilingual inscriptions in Elamite, Babylonian and the newly invented Old Persian cuneiform.

Loosely based on the signs used for Sumerian and Akkadian, Old Persian cuneiform was a far simpler system, since it followed the alphabetical principles of Aramaic. There were thirty-six signs in all – signs for the three vowels a, i and u, twenty-two signs for consonants usually linked to the vowel a, four linked to the vowel i and seven to the vowel u. Two simple signs were used as word dividers, which was to prove a valuable aid to decipherment, and single signs represented the words king, land, earth, god and Ahuramazda, as well as numerals. Unlike other types of cuneiform, the invented Old Persian cuneiform is rarely found on clay tablets, but normally as inscriptions on rock faces, metal plaques, vases, stone buildings and stone monuments. Old Persian cuneiform was in use for less than two centuries, having been abandoned by the time the Macedonian Greek conqueror Alexander the Great defeated Darius III in 333 BC, overran the Persian Empire and sacked Persepolis.

After Alexander the Great, the Persian Empire came under Hellenistic Greek rule, until it was conquered a century later by the Parthians, nomads from central Asia, around 238 BC. The Parthians and their empire survived for more than four centuries, before being overthrown in AD 224 by the Sasanians under their first king Ardashir. It was the son of Ardashir, King Shapur I, who built the city of Bishapur that Rawlinson visited in 1834 while staying at Shiraz. The Sasanian Empire also lasted over four centuries until the Islamic conquest of Persia in AD 651.

Just before his trip to Bishapur, Rawlinson had visited the ruins of Persepolis and seen for the first time trilingual cuneiform inscriptions carved in the three scripts of Old Persian, Babylonian and Elamite, but at this stage he had no idea of their significance. The obsession of the Persian kings for inscriptions carved in three languages on rock faces and buildings ensured that those inscriptions remained visible to early European travellers, whereas most inscriptions in Mesopotamia were hidden from view, awaiting discovery in archaeological excavations. Because Europeans had greater contact with Persia at an earlier date than with Mesopotamia, it was inevitable that attempts at deciphering cuneiform began here, most notably at the ruins of Persepolis. Old Persian became crucial in understanding all other cuneiform scripts, and when Rawlinson achieved his first breakthrough in the decipherment of cuneiform, it was Old Persian that was to provide the first clues.

Five: Discovering Darius (#ulink_dd6a11ad-51e3-53c3-8f3e-24a45a557fe7)

Over two centuries before Rawlinson arrived in Persia, European courts had been attempting to establish trading links with Shah Abbas I, and while on a mission to the Persian court on behalf of Philip III of Spain and Portugal, Antonio de Gouvea visited Persepolis in 1602. He found the writing of the inscriptions very strange: ‘there is no one who can understand it, because the characters are neither Persian, Arabic, Armenian, nor Hebrew, the languages now in use in the district.’

(#litres_trial_promo) Back at the Spanish court, Antonio de Gouvea met Don Garcia Silva Figueroa, who was himself inspired to visit the site in 1618 when in Persia as ambassador. Don Garcia was the first person to identify the ruins as Persepolis, but was equally puzzled by the sight of the inscriptions: ‘The letters themselves are neither Chaldaean, nor Hebrew, nor Greeke nor Arabike, nor of any other Nation which was ever found of old, or at this day to be extant. They are all three cornered, but somewhat long, of the forme of a Pyramide, or such a little Obeliske as I have set in the margin.’

(#litres_trial_promo)

Pietro della Valle, a traveller from Rome, became acquainted with Don Garcia at Isfahan, and when della Valle left the city towards the end of 1621, he spent two days at Persepolis and the neighbouring ruins. Like Don Garcia, he noted: ‘One cannot tell in what language or letters these inscriptions are written, because the characters are unknown.’

(#litres_trial_promo) He was unsure of the direction of writing, ‘whether the characters are written from right to left as is the Oriental custom, or the opposite, from left to right as with us’,

(#litres_trial_promo) but because of the way the signs were constructed, he correctly deduced from left to right. He copied five of the most common cuneiform signs, and even though publication of the account of his journey was delayed until 1658, these were still the first cuneiform signs ever to be published.

Dutch, English, French and German travellers and artists were in turn drawn to Persepolis, many publishing engravings of the inscriptions, although these were often inaccurate. Samuel Flower, an agent of the East India Company, copied several trilingual inscriptions, from which a random selection of twenty-three cuneiform signs from the three scripts, each one divided by a full-stop, was published in 1693. Scholars were misled into believing that this random selection represented a full inscription with punctuation, a confusion that lasted for over one hundred years. Thomas Hyde, a Professor of Hebrew and Arabic at Oxford, studied Flower’s so-called inscription and wrote in 1700 that the signs were purely decorative and could not represent writing, not least because they were all different and divided by full-stops. Although led astray by this composite inscription, he described the signs as ductuli pyramidales seu cuneiformes, so coining the term ‘cuneiform’.

The next significant advance was by Carsten Niebuhr, a Danish scholar and explorer who spent several days at Persepolis in 1765. When the account of his travels was published a few years later, his drawings of inscriptions at last provided reliable material for scholars to use in their attempts at decipherment. Niebuhr was the first to realize that the inscriptions comprised three different scripts and therefore probably three different languages. He misleadingly referred to each script as an alphabet, although only the Old Persian was an alphabet. Referring to the scripts as classes I, II and III, he thought that class I (recognized later as Old Persian) was more simple than the other two and had forty-two alphabetical signs (there are actually even fewer – thirty-six). He confirmed della Valle’s view that the writing was done from left to right, after observing that in two identical inscriptions the line endings were not in the same position. It did not seem to occur to Niebuhr that the three different scripts each reproduced the identical text.

Two decades later, in 1798, Oluf Gerhard Tychsen, a Professor of Oriental Languages at Rostock, published a paper wrongly alleging that the inscriptions at Persepolis were of Parthian date and claiming to make out the name of the first Parthian King, Arsaces (‘Aksak’), in a recurring group of signs. He did correctly identify one diagonal sign

in the Old Persian script as a word divider, although he was incorrect to claim it could also signify the conjunction ‘and’.

That same year Frederik Münter, a Professor of Theology at Copenhagen, read two papers to the Royal Danish Society of Sciences that were published in 1800 (and translated into German in 1802). In his opinion the ruins and inscriptions could only belong to the Achaemenid kings, and among his many observations on the inscriptions he correctly concluded that the diagonal sign was only ever a word divider. He rejected Tychsen’s reading of the name Aksak and rightly suggested that it might be a Persian title, such as ‘king of kings’. However, he incorrectly suggested that the languages of the three scripts were Avestan (or Zend as it was then called), Pahlavi and Parsi.

The known languages of Persia are divided into Old Iranian (in use up to Alexander the Great’s conquest), Middle Iranian (used up to the Islamic conquest) and New Iranian. The Iranian language belongs to the Indo-Iranian branch of Indo-European languages, and around 1000 BC Iranian speakers spread from central Asia into Afghanistan and Persia (Iran). Old Iranian comprised two known languages, Old Persian and Avestan, and the former was the everyday speech of the Achaemenid kings and was probably the spoken language of south-west Persia. It is represented by a limited number of cuneiform inscriptions, as at Bisitun and Persepolis. At the same time, Avestan was spoken in north-east Persia and became the language used to compose the Avesta or Zend-Avesta, the holy texts of Zoroastrianism that became the state religion from the time of Cyrus the Great. The Avesta was originally handed down from one generation to another by word of mouth, but was written down for the first time in the Sasanian era, probably in the sixth century AD, long after cuneiform had gone out of use. The earliest surviving manuscripts date from the thirteenth century, just as most Latin texts from the Roman Empire survive only as medieval copies.

In 1771 Abraham-Hyacinthe Anquetil Duperron published the first translation, into French, of the Zend-Avesta. Because Avestan was similar to Old Persian, knowledge of this language was to prove invaluable as it enabled decipherers to work out the vocabulary of Old Persian. Münter at Copenhagen tried to compare the frequency of cuneiform signs in the Old Persian inscriptions with the frequency of letters in Avestan texts, and although his method did not succeed, it led him to suggest correctly that the other two scripts in the trilingual inscriptions from Persepolis were translations of the first.

In July 1802, Rafaello Fiorillo, secretary of the Imperial Library at Göttingen in Germany, was out walking with the twenty-seven-year-old schoolteacher Georg Friedrich Grotefend, a man ‘possessed of an extraordinary memory and excellent health, which allowed him to study from the earliest morning until late at night, without stint or relaxation … although he was considered by persons not in his intimacy, to be of a cold and reserved character, wholly occupied with his recondite studies, and uninterested in anything beyond them, this learned man was really full of feeling, and endowed with an almost child-like simplicity, which endeared him to all those who were of the circle of his friends.’

(#litres_trial_promo) Fiorillo asked if it could ever be possible to understand cuneiform inscriptions, when both the alphabet and the language were absolutely unknown. Rising to the challenge, Grotefend chose two trilingual inscriptions copied by Niebuhr at Persepolis that looked very similar, examining in each the most simple of the three scripts (the Old Persian). Just a few years earlier, Silvestre de Sacy, an Oriental scholar in Paris, had worked out the meaning of Sasanian (Middle Persian) inscriptions from Naqsh-i Rustam near Persepolis, using the aid of identical inscriptions in ancient Greek. Because they contained the names and titles of kings, Grotefend thought that his cuneiform inscriptions from Persepolis might be similar, but perhaps dating to the time of Xerxes. From these suppositions, he thought that the inscriptions would include the formula ‘Xerxes, great king, king of kings, son of Darius, great king, king of kings, son of Hystaspes’.

Using this method, Grotefend successfully identified the groups of signs for Xerxes, Darius and Hystaspes and also the group of signs for ‘king’, but did not work out all the individual cuneiform signs correctly. Using versions of the names derived from ancient Greek, Hebrew and Avestan, he thought that the cuneiform signs

for Darius represented d-a-r-h-e-u-sh (darheush), although in fact they spell da-a-ra-ya-va-u-sha (darayavaush). The cuneiform signs for Xerxes