Quick viewing(Text Mode)

The Alphabet Tree Dren’S Puzzles

The Alphabet Tree Dren’S Puzzles

TUGboat, Volume 26 (2005), No. 3 199

A major intellectual step was the invention of the rebus device. This is where the sounds cor- responding to are combined to form . have all come across these, often as chil- The tree dren’ puzzles. For example, a picture of a bee Peter Wilson plus a picture of a tray can represent the word ‘be- tray’, or more obscurely a picture of a bee plus a 1 Introduction picure of a female deer can stand for the word ‘be- We got from via Akn and ’ to hind’. Even a single can suffice; for in- XPD stance a pictogram of the sun can be used for both ` k p A K N, and and @ ¼ à in only about 2000 the ‘sun’ and ‘son’. Consequently years. This is a short story of how that happened can be created that just stand for sounds and then and how we know what the strange symbols mean, they can be combined to form words. This can a k n a k n .., and . The in all the markedly reduce the number of symbols required for examples were designed for use with TEX, especially a full . Symbols representing sounds by humanities scholars, and are freely available. are called phonograms. Phonetic writing requires 1.1 Writing systems few . In these writing systems, the glyphs represent sounds. All writing systems are a com- The earliest known writing is from Sumeria dating bination of phonetic and logographic elements but from about 3400 bc. In some other places the earli- the proportions of these two elements vary among est discovered date from: in 3000 bc, . Ideographic scripts essentially have one the Indus Valley in 2500 bc, in 1900 bc, for each word, and this usually represents the in 1200 bc and Central America in 600 bc. These meaning of the word, not its pronunciation. The all looked very different. numerals 1, 2, . . . are pronounced in English It seems that initially writing was used for bu- as one, two, . . . but in German as eins, zwei, . . . even reaucratic purposes — keeping accounts, recording though the meanings are identical. Mixed systems goods and so on — and a limited was are where some are ideographic and others are sufficient for this. Later, to record the great deeds phonetic. For example, in English 1st, 2nd, . . . , of the rulers and especially their names, to promul- are with a phonetic component so that gate their laws, and to meet the needs of the reli- we read them as first, second, . . . instead of onest, gious establishment, writing would be extended to twond, and so on. a full writing system: a system of graphic symbols, Although it does have some phonetics the - or glyphs, that could be used to convey any idea. nese is principally logographic; this may be At that point literature became a possibility. All because of the Chinese itself. Spoken Chi- writing systems represent speech in one form or an- nese consists almost entirely of one- words other. Some glyphs represent sounds while others and there is a limit on the number of short sounds are semantic signs that represent either words or 1 that the can make; the Chinese use concepts; these are called . something like 400–900 sounds. Many sounds, there- It appears that early writing systems followed fore, have shared meanings — . In spo- the same general progression. The first actual writ- ken English the meanings are deduced from the con- ing was pictographic or iconographic where a simple tex and in writing by their . For example: picture designated a real object — a drawing of a ‘Pare me a pair of pears’. In spoken Chinese homo- deer represented a real deer, for example. Gener- phones are partly distinguished by using four lev- ally the pictures were very simple and abstractions els of pitch (thus increasing the number of different of what we might think of as a drawing. A stylised word sounds to some 3000), and by context. In writ- picture is called a pictogram. ing there is little possibility of reducing the number Gradually the pictures were formalized and also of glyphs required to represent the vocabulary from began to be used to represent relationships and ideas that of an ideographic script. Chinese and Japanese as as objects. This is called ideographic writing. use the same ideographic script although their lan- For example, a picture of the moon could represent guages are very different. the idea of night or darkness as well as that of the At the other end of the spectrum, the Finnish moon itself. A standing for an idea is a and French systems are much closer to pure pho- semantic and is called an . netics, with some logographs; that is, the phonetic values of the set of glyphs is closely matched to the 1 From the Greek = word. 200 TUGboat, Volume 26 (2005), No. 3 sound of the . In the English writ- soon after and was initially the Egyp- ing system the alphabetic glyphs, a– and A–Z, rep- everyday business script. The demotic3 script resent sounds, the ‘?’ mark represents an idea and came much later, around 650 bc, and was then used glyphs like ‘$’ represent whole words (which can also as the everyday script. The priests, though, contin- be spelled out as ‘dollar’). ued to use the script. From the several thousand characters that an Unlike hieroglyphics, which failed to spead be- educated Chinese or Japanese needs to know, the yond Egypt, writing became popular and English speaker only needs to know 26 (52 if you in- was taken over by the Babylonians when they con- clude both uppercase and lowercase). Monolingual quered the Sumerians in 1720 bc. Chinese and Japanese can read and understand each There is evidence that there was a writing sys- other’s scripts because they are based on the same tem in Crete about 3000 bc. A thousand years later ideographic system, even if they can’ understand a syllabic script called was in use, and each other when speaking. Monolingual French and by about 1800 bc this had been replaced by the English for example, with their alphabetical writing Linear syllabic script which was used for writing systems, are not in this happy situation; they can Mycenean Greek. read each other’s scripts but with no understanding By about 1500 bc, in- of either the written or spoken words. cluded ‘alphabetic’ glyphs alongside all the others. Phonetic scripts can be roughly classified into The alphabetic cuneiform script dates from two major kinds. about 1300 bc. • In a syllabic system a glyph represents a syl- 1.2 The alphabet lable, usually a followed by a (CV) but can be a VC pair or a consonant vowel Although it may never be possible to describe ac- consonant (CVC) triple. curately the origins of the alphabet, scholars are generally agreed that most of the world’s • In an alphabetic system a glyph represents ei- are descended from one that was probably invented ther a vowel or a consonant, again with two sub- about 1600 in the . This was a divisions. In some alphabetic systems, like - bc Semitic alphabet, where Semitic refers to a linguistic brew, only the are denoted, whereas family that ranged between the Sinai in the south, in a full alphabetic system, like French, both along the Mediterranean coast, north through consonants and are fully represented. Minor and east to the Euphrates valley; the Canaan- By about 3200 bc the Sumerians were using a ites and Phoenicians, among many others, spoke a cuneiform (wedge-shaped) script. They lived in the Semitic language. in the area of the Tigris and Eu- As people travelled, particularly as conquerors phrates in what is now the Middle East. The Sume- or merchants, the original alphabet was dissemi- rians had over 2000 ideograms. Following their dis- nated geographically and gave rise to several - covery of the rebus device they eventually reduced betic branches. Roughly speaking, in the time - the number of glyphs in their script to about 600, riod 1400–1100 bc, these were Archaic Greek; Old which included both semantic glyphs and phono- Hebrew which led to Samaritan; Phoenician; and grams. The Sumerians wrote on clay tablets using South Arabic scripts which in turn led to , a stylus to impress the marks. Drawing on clay is via Ethiopic, and other obscure scripts like Thamu- not easy, which must have given them an impetus to dic and Lihyanic. With only minor exceptions these move away from pictograms towards ideograms and scripts were used for writing languages from the then on to phonograms. Indo-European family. All are interesting in their 2 The Egyptian hieroglyphic writing system de- own right, but the most relevant script for us is the veloped roughly in parallel with the Sumerian sys- Phoenician which is the direct ancestor of not only tem. The , though, wrote on with our modern alphabet but also of many other a reed brush or pen, or painted on the walls of tombs. alphabets and scripts in use today. Figure 1 shows Drawing with these implements is much easier than the main descendants of the . scratching pictures in clay and thus they did not At the earliest time there was no fixed direction have such a great need to move towards phonograms. to writing. It could be left to right or right to left, For everyday purposes they did develop more effi- randomly, or at times lines would alternate between cient writing methods with their hieratic and left to right and then right to left. This alternation scripts. The hieratic script was invented

2 From the Greek for sacred engraved writing. 3 From the Greek ‘demotikos’ meaning ‘in common use’. TUGboat, Volume 26 (2005), No. 3 201

Proto-Semitic Hieroglyphs

Phoenician 1100 bc @ @ @ Greek 850 bc 1000 bc aXX @aaXX aXXX @ aa XXX @ aa XX Nabataean Armenian Indian Etruscan 350 bc Hebrew ¢B ? ? ? ¢ B ¢ B ¢ B Roman Persian Arabic @ ? ? @ @ @ Gothic Latin Cyrillic ? ? ?

Figure 1: The alphabet tree of writing direction is termed boustrophedon4 writ- sound values; only later may new characters be in- ing. Typically, in writing, two sets of troduced. However, it is more likely that existing glyphs were used, one being the image of the characters will be decorated to denote new sounds. other, so those who were literate would have twice For example, French and English use the same set as many characters to remember. of alphabetic characters, but the French add accents to some characters (for instance, `e, ´e,. . . ) to denote 2 Changes and sounds used in spoken French but not in English. As we will see, scripts are remarkably resistant to Thus, a particular script may be used for sev- change, but nevertheless they do. There are two eral languages, and over a period of time one lan- main reasons why a script should change. guage may be captured using several scripts. As the One is a change in , which during ages pass civilisations also pass, and their scripts the period considered comes down to the writing may be forgotten until rediscovered by archaeolo- materials used. These ranged from using pointed gists, by which time their languages may also have implements to make marks in soft clay, reed brushes been lost. According to Robinson [Rob02, p. 262] to paint or write on a smooth surface, and scratching decipherment ‘is a process of deducing from texts or chiseling letters into hard stone. a known or plausibly reconstructed language that The other reason is a script starting to be used accounts for the patterns of sign use in texts.’ A to denote a language that it was not designed for. decipherer is faced with three possibilities: Every script represents the sounds of a language, • A known script, in the sense that the ‘meanings’ and not every language has the same set of sounds. of the glyphs are understood, and an unknown It is easy enough to drop characters that do not language; represent a sound in a language but it seems much harder to introduce new characters for new sounds. • an unknown script and a known language; or, If there are redundant characters they are given new worst of all,

4 From the Greek for ‘as an ox plows a field’. • both the script and language are unknown. 202 TUGboat, Volume 26 (2005), No. 3

A few very brief descriptions of some decipher- Just to help you along, the scripts are all writ- ments are given later. In case you would like to ten left to right, even if the original scripts were understand better some of the problems of decipher- written in different directions, and are more or less ment, you can try your hand at the following. of the following English paragraph. Tis is a sixT gentry greek sgript This is ——. When the quick brown fox yhen qyik broyn Fox ymped oyer jumped over the lazy dog it was time for all Te lazy red dog it yas time for all good men to come to the aid of the party. good men to gome to Te aid of Te ——. party Tis is fairly simple In real life the task is much harder than the this: is: : ugaritig: examples imply. A representative set of example guniphrm: alphabt: texts has to be gathered and then two things have to whn: th: quik: brwn: be done before the hard work starts — determine the phH: Dumpd: phr: th: writing direction, and determine the kind of script. laZy: rd: dg: it: was: The writing direction is often determinable by tim: phr: all: gd: mn: T: noting whether the text is ‘set’ ragged-right (left to gum: T: th: aid: ph: th: right direction) or ragged-left (right to left direc- party: th: answr: my: tion). Of course some texts may be ‘centered’ or phrnd: ‘justified’, which is not much assistance. The other preliminary task is to decide on the kind of script. Scripts can be classified into one of this is th heglyphik skript three types: phonetic, ideographic, or mixed. In a whn th kwik brwn phC phonetic script the glyphs represent the sounds of Dumpd phr th lazy rd dg the language. Alphabetic scripts aim at the ideal of one sign per sound, but this ideal is rarely met. it ws tim pq all gd mn t gm Syllabic scripts use one sign for each syllable. The t th aid ph th party why th kind of script can usually be determined by count- smil n th phaz ph th sphinks ing the number of different glyphs. An alphabetic script typically has between 20 and 30 characters, a 3s is T inr YrpO Wn T v6c pVn has roughly 30 to 60 characters and more oks bmPd oW T aZ Rf gc iT ws OM than that indicates ideographic elements. However, or a vd Mn 4 hM 4 T aD T pr3 if you apply that naively to English writing then is anYent kRk 4 it looks like a syllabary as there are 52 alphabetic ths s th prt smtk aphabt whn characters (26 in each case) plus numerals and punc- th qwk brwn phks mpd phr th tuation. The same problem may arise with ancient lazy rd dg t was tm phr all gd scripts, as it is not always easy to decide whether mn t gm t th ayd ph th prty rly two similar looking characters are the same or not. sgrpt 1 from [Coe99] lists the numbers of glyphs in some different kinds of scripts. ths s th phnSn aphabt whn th qwk There is an empirical formula for estimating the brwn phks mpd phr th lazy rd dg t was probable number of signs in a script from a small tm phr all gd mn t gm t th ayd ph th sample of the script [Rob02, p. 310]. It seems to prty ltr wrtng work for modern languages and scripts such as Ara- 3s is T YIO5 YlpO Wn T vwc bic, English, and Japanese as well as ancient prn jmP ow T l9 Rg Hg it ones such as . ws 3M or al g5 Mn 4 gM 4 T In a small sample of an alphabetic or syllabic aT T pr3 is gR 4 M writing system consisting of a total of characters of K different kinds then the probable number of sym- : is: o: pSn: slbr: bols S forming the alphabet or syllabary is, subject hn: o: kik: bn: fx: to various restrictions not enumerated here, given jmpd: : o: lz: RP: approximately by the formula dg: it: vs: tm: : 2 al: GP: mn: T: Gm: S = L /(L − K) − L. (1) T: o: aD: f: o: prT: Applying this formula to the previous paragraph is: bly: iN: o: VnP: where, ignoring the the formula itself, uppercase TUGboat, Volume 26 (2005), No. 3 203

Table 1: Numbers of individual glyphs in writing systems Logographic Syllabic Alphabetical Sumerian 600(+) Persian 40 English 26 Egyptian 800 Linear B 87 Anglo-Saxon 31 Hittite 497 Cypriot 65 35 Chinese 5,000(+) Cherokee 85 Hebrew 22

3.1 Sumerian and Ugaritic Table 2: Common transliterations and their pronunciation The earliest script so far discovered is Sumerian cu- neiform dating from about 3200 bc, which had de- Symbol Pronunciation veloped from earlier pictograms. At its most bloated d, as in ‘did’ it included over 2000 glyphs but as it proceeded d dj, like in ‘joke’ or di in the French ‘dieu’ ¯ through the normal evolutionary process the num- g hard g, as in ‘get’ ber of glyphs dropped to about 600 in its final form. h, as in ‘home’ The Ugaritic cuneiform script dates from about h an emphatic h, sounded in the throat . 1300 bc and was alphabetical, although like most h , as in the Scots ‘loch’ scripts for the did not include ˘h softer than h, like the ch in German ‘ich’ ¯¸ ı , as in ‘yea’˘ vowels. The script consisted of 30 letters and a k k, as in ‘kit’ ideographic (a short vertical wedge). k. k in the back of the throat, like the Arabic It appeared to have got the alphabetical idea from in Qur‘ˆan (Koran) contemporary linear scripts. It was used to write a trilled r, as in Scots ‘rain’ a language related to Hebrew. In addition to the s s, as in ‘soap’ typical administrative texts there are a number of ˇs , as in ‘ship’ mythological texts about the god Baal, which give t t, as in ‘tub’ scholars another view on some Biblical stories. .t t, as in ‘tune’ The order of the glyphs in an alphabetic script t tj ¯ is usually revealed by writing found from scribal w, as in ‘wet’ schools where the pupils were practising their abe- y y, as in ‘yes’ z or ’ , like the break in the Cock- cedaries. In the original order, which is reasonably ney pronunciation of ‘bottle’ as ‘bo’’ or typical of Semitic scripts and with the word divider the American pronunciation of ‘Seattle’ as (:) being the last glyph, the and ‘Sea’el’ the modern is: ‘ gutteral, the Semitic ’ b g d h w z H ’ b g h d h w z h. ˘ T y k X l m D n Z characters and , there are 232 charac- t y k ´s l m d n z . ¯ . ters of 24 different kinds (q and z are not used) we s ‘ p x q r J G t get the approximate value of s ‘ p s q r t ˙g t . ¯ S = 2322/(232 − 24) − 232 = 26.77 i V : i u `s : for the number of signs in the lowercase English al- The last recorded use of a cuneiform script was phabet compared to the actual value of 26 signs. in 75 ad, so cuneiform vies with hieroglyphs for the longest period of use of any script.

3 The earliest scripts 3.2 Hieroglyphs We now show a of scripts dating from before Hieroglyphs were used by the Egyptians from about 1000 bc, some of which are related. Transliterations 3000 bc to 400 ad. The script is a mixture of a set into modern Western characters are also given; Ta- of consonantal glyphs, a syllabary, and logograms. ble 2 lists the main transliterations used and their pronunciation. 204 TUGboat, Volume 26 (2005), No. 3

Table 3: A sampler Glyph Sound Meaning Glyph Sound Meaning

‘ arm z vulture A a b b leg B b h ball of string? ˘d cobra d hand D ¯ d ¸ ˇsz pool with flowers ır eye E e f horned viper F f g jar stand g h. twisted wick h H h ¸ eat, drink, speak ı reed I i house, building j k slope of hill k with handle K .¸ k ıw walk, run l lion L l m pair of ribs? m owl M m n water n palm of hand p P p tp head hr face Q q . wr small, bad, weak r mouth R r s door bolt s folded cloth S s t tethering t T ¯ t w w quail chick W w h ox door X ¯ x rejoice y pair of reeds Y y ˇs stone, pool ¸ z ımy crossed planks awt shepherd’s crook + ? kmz throw stick / .

There are approximately 6000 known different hi- Egyptian town called Rosetta, by French soldiers in eroglyphs, but fewer than 1000 were in use at any Napoleon’s invading army. The stone carries an in- one time. scription in three different scripts: hieroglyphs at A short sample of hieroglyphs is shown in Ta- the top, which was badly damaged with about half ble 3. As an example of Egyptian writing, the fol- missing; Egyptian demotic script in the middle; and lowing hieroglyphs: Greek at the bottom. There are 54 lines of Greek with the right hand ends of the last half being dam- qR t n G UwLJf | ryNf t cmv | fZwV aged or missing. The demotic portion has 32 lines, are transliterated as: ¸ written right to left, and the right hand ends of the wd hm.f hr wrryt.f nt d ‘m ıb.f zw first 14 are damaged. The first half of the lines of hi- ¯ . . ¯ and can be translated as: eroglyphs are completely missing and the existing 14 His Majesty departed upon his chariot of electrum, lines, which correspond to the last 28 lines of Greek, his heart joyful. are damaged at both ends. The is now kept at the . There were also hieroglyphs for numerals, some The first attempts at decipherment focused on examples being: (1), (2), (10), (100), | ¸ 2 3 the demotic script. Initial partial (1,000), . . . (1,000,000). 4 7 were accomplished by the Frenchman Sylvestre de The breakthrough in the decipherment of hiero- Sacy (1758–1832) and the Swedish diplomat Johan glyphics came after the Rosetta Stone was discov- Akerblad˚ (1763–1819). The basis was being able to ered in July 1799 near Rashid, which was the ancient TUGboat, Volume 26 (2005), No. 3 205

n p l t toMys rB y gaO Hieroglyph Young’s value Hieroglyph Young’s value p bir p B t e t r not essential n o n lo or ole i l y ma or m superfluous M g i ke or ken y a osh or os s t feminine termination O Figure 2: Young’s decipherment of the and Berenice identify corresponding names, such as Alexander, he was probably also bored by the time he got to Ptolemy, and Berenice, in the Greek and demotic this stage. texts. This gave sounds for some of the demotic The final decipherment was achieved by the signs and from this it was possible to show that the Frenchman Jean-Fran¸coisChampollion (1790–1832), script had phonetic components. From the identified who was more open minded than his predecessors. signs it was possible to identify some other words At age ten, on having been shown hieroglyphs by the such as temple and . Unfortunately, Akerblad˚ French mathematician Jean-Baptiste Fourier and was convinced that the script was entirely phonetic, being told that nobody could read the strange writ- which blocked any further progress on his part. ing, he decided that he would solve the mystery. The English polymath Thomas Young5 (1773– To equip himself for the task he studied Arabic, 1829) then took up the challenge in 1814. Young Chaldean, Chinese, Coptic, Ethiopic, Greek, He- was a prodigy; he could read fluently before he was brew, Latin, Pehlevi, Persian, Sanskrit, Syrian, and three, and by the time he was fourteen he had stud- Zend. By 1822, through systematic analysis of the ied Arabic, Chaldean, Ethiopic, French, Greek, He- available material, he showed that the hieroglyphic brew, Italian, Latin, Persian, Samaritan, Syriac, and script had phonetic principles. To progress further Turkish. He proved that the demotic and hiero- he needed to have two or more known names with glyphic scripts were not completely distinct and that some hieroglyphs in common so that they could act the Egyptians used a mixed writing system. He was as a check on any proposed decipherment. able to decipher much more of the demotic and - In 1819 W. J. Bankes had had an obelisk moved tablished the equivalence of many demotic and hi- from Egypt to his home at Kingston Lacy in Dorset, eroglyph signs. He determined that the only royal England. The hieroglyphs included two different name appearing in the hieroglyph section was Ptol- cartouches and the Greek inscription at the base of emy. This was spelt phonetically in demotic and he the obelisk mentioned Ptolemy and . He surmised that it was also spelt phonetically in hi- noticed that one of the cartouches was identical to eroglyphs, corresponding to the Greek (Ptolemaios). that deciphered as Ptolemy by Young, and surmised Young produced the list of values given in Figure 2. that the other corresponded to Cleopatra. Bankes From an inscription at the temple of Karnak had lithographs made of the inscriptions, annotated he also had the name of the queen Berenice (Greek with his idea about the cartouches, and distributed Birenike) and for this he constructed the further cor- them in 1821. When Champollion received a copy respondences, also shown in Figure 2. he made the decipherment shown in Figure 3. This is about as far as he could get, as he be- There was a remarkable degree of similarity be- lieved that the vast majority of hieroglyphs were tween the values from the two names, except for the ideographic, and phonetic spelling was limited to and signs which he explained as being homo- t d the names of foreigners. As his mind was so quick phones — they could each represent the same sound (t in this case).

5 If you have taken any science courses you have probably heard of Young’s Modulus and Young’s Rings. 206 TUGboat, Volume 26 (2005), No. 3

p l K d toMys liopara Hieroglyph Champollion’s value Hieroglyph Champollion’s value p c p K t l t l o e o i l o l o m p M p e a y a s t s d r r a a Figure 3: Champollion’s decipherment of Bankes’ Ptolemy and Cleopatra cartouches

He then looked at other cartouches to see if he could generate recognisable names from them by ap- ŢŃĽ plying these sound values. The first one he tried was: J u c b b 8 E f k t b f W Ţ Ń ´ 1 d d R b D Ţ Ń ˆ 2  l nr aksidS p D Ţ Ń ´ 1 p Y T S Ţ ´ 1 He was able to spell this out as al??tr?, which Q r Y j Ţ Ń ´ 1 seemed to match the Greek alksentrs (Alexander), ayCpYpJSŢŃˆ 2 thus giving him three more sign values. Further e O F Ţ Ń Ľ ˜ 3 cartouches both confirmed his values and gave new " d D Ţ Ń Ľ ´ 1 ones. One nagging thought was that only foreign a M b J R j Ţ Ń Ľ ¨ 4 names might be spelt phonetically but, among oth- 4 1 Ţ ˜ Ń ˆ Ľ ˆ 3 2 2 ers, this  showed that this was ˆ´s Figure 4: Example of a Linear B text and partial not the case. Champollion knew that the ibis, ,  ˆ interpretation was the symbol of the god and he read the cartouche as Thoth-mes, an old Egyptian name. As he matched more signs with sound values he was increasingly able to read the hieroglyphic texts as well as the names in the cartouches, and eventu- ally could identify the as Coptic. kinds of animals, such as horses and pigs, and for Champollion’s work laid the foundation for Egyp- trade goods, such as pots or wool. tology as it is known today. Clay tablets bearing the script were found by Sir Arthur Evans (1851–1941) while excavating the 3.3 Linear B ruined Minoan palace at Knossos in Crete, starting The Linear B script was a syllabary that was used in 1900. The tablets were usually small enough to be during the period approximately between 1600 and held in the hand, the largest being about six inches 1200 bc. Most of the examples come from Crete, across, ten high and an inch thick. The tablets were particularly Knossos, but there are some from the accountancy records of some kind and he did work Greek mainland. out their numeric systems but not much more than The script consists of some 60 basic signs, 16 that. As an example, Figure 4 shows, on the left optional signs, and about 11 signs that have yet to side, the text of a fairly typical tablet. On the right be deciphered. The script also had signs for numbers side is an interpretation of Ţ, Ń, and Ľ, which look (1–1000), and signs for various kinds of weights and rather like an addition sum, where Ţ, Ń, and Ľ measures. There were also sets of signs for different might be units in a non-metric system, like fluid TUGboat, Volume 26 (2005), No. 3 207

Word A Word B in three forms, as illustrated in Figure 5, which be- came known as ‘Kober’s triplets’. ;c3j hEYj Case 1 Ventris had done his own analysis of the script Case 2 ;c3b hEYb and had come to the conclusion that it was a syl- Case 3 ;c4 hE1 labary. He argued that the difference between the words 41 or 4s for total might be due to gender Figure 5: Two of Alice Kober’s triplets differences in an inflectional language as the first form occurs with the ideogram for man, and the ounces, pints and gallons.6 With a modicum of - other with the ideogram for . If this were the 1 s fort it can be shown that Ţ = 3 Ń = 18 Ľ. case then the consonants in and were probably There were many tablets like these where the the same but the vowels were different. By analysing last line started with either 41 or 4s, and it was a number of words in this way he was able to start reasonable to assume that these words meant some- building up a grid where the signs in each row had thing like ‘total’. the same consonant and those in the same column On one tablet to do with listings of horses Evans had the same vowel. It was still a long road, though, noticed a pair of signs, HU, which matched the from having the signs coordinated in this fashion to Cypriot signs Hf po-lo (see Table 8), sim- being able to read them. ilar to the Greek polos for foal. He was convinced From Kober’s work he noticed that there were that the Cretans spoke an unknown language, which groups on the Knossos tablets that were not on the he called Minoan, a theory that he held throughout Pylos tablets, and made a bold leap to thinking that his life, going to great lengths to disparage anyone they might be the names of places on Crete. He hE1 who did not agree. Evans guarded his finds some- suggested that the signs might be the word ayC1 what jealously and made little publicly available for for the Greek Knosos (Knossos) and could others to work on. It was not until 1952, well af- be the word for the Greek Amnisos, the port for ter his death, that descriptions of his tablets were Knossos, and a few other names. When he applied published. the guesstimated values to the signs in the grid he However, another trove of Linear B tablets had was able to assign values to other signs and start 41 4s been unearthed at Pylos, on the Greek mainland, by ‘reading’ a few things. For example, and the American Carl W. Blegen in 1939. These were read to-so and to-sa, which were similar to the Greek published in 1951 and would probably have been tosos (masculine) and tosa (feminine) for ‘so much’ available earlier if the world had not been consumed or ‘so many’. Ventris had originally whole-heartedly with the other events of 1939 and later. agreed with Evans that the language of Linear B was Michael Ventris, a British architect, had been not Greek, but it was now appearing as though it fascinated by Linear B since he was a schoolboy and might well be, especially if the Cypriot polos clue devoted much of his spare time in trying to deci- was included. pher it. Initially there was no success because of the Ventris was familiar with the Greek of paucity of material to work on, but the publication (about the ninth century bc) from school but the of the Pylos tablets changed that. Linear B tablets were much older than that. The In the meantime the script had been analysed words he was reading seemed similar to Homeric by various scholars, the signary had been estab- Greek but were not the same. For example there lished, and lists had been made of which signs were were several tablets from Pylos listing numbers of most common at the start and end of words, and women, from the ideograph, often followed by two hw h7 of how signs tended to group themselves. Dr. Al- other words, and , also with numbers and ice E. Kober (1907–1950), a classicist at Brooklyn it was reasonable to assume that these words might College, had noticed groups of signs where all but be equivalent to ‘boys’ and ‘girls’. However he read the last one or two signs in a word were the same, them as ko-wa and ko-wo, whereas the Greek that and thought that this might mean that the lan- he knew was kourai and kouroi. In general the Lin- guage captured by Linear B was inflectional, like ear B appeared incomplete, and even when Latin or occasionally English as in ‘I write’ but ‘he filled out only close to . writes’. In particular she noted words that appeared At this point he formed a partnership with John Chadwick, a lecturer in at Cambridge Uni- 6 The British and the Americans agree that there are eight versity whose speciality was the early history of the pints to one gallon, but, perhaps to the chagrin of Texans, . Together they worked out a consis- there are 20 fluid onces to a British pint and only 16 to an American pint. tent set of rules describing how Greek had changed 208 TUGboat, Volume 26 (2005), No. 3

Table 4: The basic Linear B syllabary Table 5: The Proto-Semitic signary a e i o u Name Glyph Sound Meaning a e i o u alpu aA ’ ox d d D f g x betu bB b house g j j J b g throw stick? dD d fish? k k K c h v z z m M y A B m eE h man? n n N C E F wawwu w w , peg p p P G H I hotu hH h. fence q q Q X 8 T .t? twisted flax r r R O U V yadu yY y hand, arm kK s s S Y 1 2 kappu k palm of hand lamdu l ox t t T 3 4 5 mayyuma? m m water w W 6 7 w nahasu n n snake z z Z 9 enu oO ‘ eye s s? pP p leg/foot? between Mycenaean and Homeric times. In other xX .s? plant? words, if you took an arbitrary Linear B tablet, qQ q? knot? deciphered it and then applied their rules the re- rasu rR r head sult would be Homeric Greek. In 1953 they jointly S ˇs lotus pool? summmarised their work in an article entitled ‘Ev- vV ? idence for Greek in Mycenaean archives’ in tawwu t t mark The Journal of Hellenic Studies. Their theory was completely unexpected and its reception was mixed, to say the least. However it was soon dramatically confirmed. The American excavation at Pylos had resumed is no other relationship between this ancient script in 1952 after the break for the Second World War. and the . More Linear B tablets were found and stored for later reading. In the spring of 1953 the leader of 3.4 Proto-Semitic the team, Carl Blegen, returned to armed Around 1600 bc there were alphabetic scripts in use with an advance copy of Ventris and Chadwick’s ar- in the Middle East that are variously called Proto- ticle. Among the newly found tablets was a large one Siniatic, Proto-Canaanite, etc. I have lumped these with pictures of three-legged cauldrons, pictures of a together into a Proto-Semitic . Several of the number of jars with differing numbers of loops (han- signs in this alphabet are obviously derived from dles) on top, and Linear B inscriptions. When Ble- Egyptian hieroglyphs, and it may have been a pre- gen applied Ventris’ decipherment he read tr-ri-po- cursor to the Phoenician script. de, almost identical to the Greek tripodes for three- The alphabet consisted of 23 letters, some of legged cauldron. Next to the jars with three loops he which had alternate forms. Writing was generally read ti-ri-o-we-e or ti-ri-jo-we, and by the jars with from left to right, but could be vertical or in other di- four loops qe-to-ro-we. The Greek for three in com- rections. Table 5 shows the signary, although there pounds is tri- and experts in archaic Greek could is not a complete consensus on this. accept that quetro- would be four in compounds. Once the ‘tripod’ tablet became known most schol- 4 Phoenician ars accepted that Ventris had deciphered Linear B, The Phoenicians initially wrote right to left or left to although a few die-hards even went so far as to sug- right. The alphabet consisted of 22 letters although gest that the tripod tablet had been ‘planted’ at a 23rd glyph was used as the vav (or vau) , Pylos! which had two forms: w and f. Around 1100 bc the Table 4 shows the signs in the basic syllabary. Phoenician alphabet had stabilised and the writing Although Linear B was used for writing Greek, there direction was finally fixed as right to left. TUGboat, Volume 26 (2005), No. 3 209

Table 6: Evolution of the Phoenician script Name Meaning Hiero. Proto. Phoen. Sound ox ’A ’ ’ X beth house bB b b j camel g g g / daleth door dD d d F he window? eE h h Y vav nail w f, w w u dagger? z z z fence? Hh H h x . T T t H . yod hand yY y y d palm of the hand kK k k P lamed ox goad lL l l ? water m m m n fish n n n D prop or post s s ayin eye ‘O ‘ ‘ e pe mouth ? pP p p b sade xX x .s knot? qQ q q head rR r r Q teeth S S ˇs E tav mark or t t t +

Table 6, which is somewhat speculative in some of vav became the Greek . The names of cases, illustrates how the Phoenician script (may the letters lost their meanings and instead effectively have) developed from the Proto-Semitic script; the stood for the pronunciation of the letter. The name and, where known, the meaning of the Phoeni- also added the psi, and omega characters. Sev- cian glyph is given, as is the transliterated sound eral different glyphs were used for each character, value. It also shows how some of the Proto-Semitic depending on geographical location, whether on the glyphs may have been inspired by the Egyptian hi- mainland or around the . One variety of eroglyphs; in some cases the derivation is obvious. the 6th century bc alphabet looked like this:

abgdeFzhTiklmnxopqrstyXfPO 5 Later Western scripts Among the later Western scripts — those developed In 403 bc the Athenian citizens codified the al- after 1000 bc — Greek, Etruscan and Latin are the phabet with the glyphs looking much as they do to- direct ancestors of the . Others, day. The digamma and the qoph characters were such as Cypriot and Runic, are off by themselves. dropped from the abecedary, thus leaving the 24 characters that we are now accustomed to. The 4th 5.1 Greek century bc alphabet was like this: Initially the Greeks used the Phoenician alphabet and also wrote right to left but by the abgdezhTiklmnxoprstyXfPO it became boustrophedon and around 500 bc they 5.2 Etruscan finally settled on writing left to right. The Greeks added new letters to the Phoenician abecedary so The Etruscans, forerunners of the Romans in , that around the 6th century bc their alphabet con- based their alphabet on the Greek abecedary, but sisted of 26 characters. The Y form of the Phoeni- they continued to write right to left as the Phoeni- cian vav became the Greek while the F form cians had, so their glyphs were mirrored with respect 210 TUGboat, Volume 26 (2005), No. 3

Table 7: The changing alphabets Phoenician Greek Etruscan Roman Modern 600 bc 394 bc Modern a a a A a A A b b b B b B B g g g Γ g CG CG d d d ∆ d D D h e e E e E E w y y Υ VY UVWY f F f F F z z z Z z Z Z H h h H h H H T T T Θ T y i i I i I k k k K k K K l l l Λ l L L m m m M m M M n n n N n N N s s s Ξ x X X o o o O o O O p p p Π p P P x S q q q Q Q r r r P r R R S s s Σ s S S t t t T t T T f f Φ f x x X P P Ψ P O O Ω

to the Greek ones. The Etruscan script was used un- again adding and dropping characters. They added til the first century ad, although the Etruscans had the G and Y characters but dropped theta, psi, and disappeared well before then. phi to end up with a 23 character abecedary, al- The Etruscan abecedary originally consisted of though they rarely used the H, K, and Z characters. 26 characters but by about 450 bc it had decreased So, their alphabet looked like: to only 20. One interesting glyph that was dropped looked like the digit 8, and denoted an ‘f’ sound. ABCDEFGHIKLMNOPQRSTVXYZ Some scholars have surmised that this was the an- The lettering used in this last example is a copy cestor of the mathematical sign for infinity (∞). In of the capitals engraved on the Trajan Column in the 8th century bc the alphabet looked like: which was erected in 114 ad. Many typogra- abgdeFzhTiklmnxopSqrstyXfPv phers believe that these represent the high point in As far as decipherment is concerned, Etruscan the Roman artistic legacy. Unfortunately they do is a known script used for writing an unknown lan- not reproduce well in the size shown here. In real guage. Scholars are able to read aloud the script but life the inscription is about six feet above eye level they do not understand the language. Apart from and there are six lines of text. The letters on the proper names only about a couple of dozen words top line are somewhat over four inches tall, decreas- are known, and these are mainly to do with family ing to about three inches on the lowest line, which relationships like ‘son’ and ‘father’. presents the illusion that they are all of the same height when observed from the normal viewpoint. 5.3 Latin Summarising, Table 7 shows the 1100 years of The Romans based their script on the Etruscan one, development of the . TUGboat, Volume 26 (2005), No. 3 211

It is very noticeable that the letter ordering is Table 8: The completely different from any of the other abeceda- a e i o u ries in Table 7. It is interesting to speculate whether a e i o u the ordering of an original abecedary depends on the frequency of use of the letters, or those with the most g g important meanings have priority. j j b The wen character (W) is no longer used in En- k k K c h v glish, but does indicate that the Anglo-Saxons had l l L d f q need of a ‘W’. The character (ˆ) is like theta m m M y A B in that it represents the ‘th’ sound. Early printers ˆ n n N C E F usually did not have a , so they used a ‘Y’ char- acter instead. From this practice comes the mod- p p P G H I ern affectation of naming something like ‘Ye Olde r R O U V r Pub’ instead of ‘The Old Pub’. Also, it has the ger s s S Y 1 2 character (J) which corresponds to the modern ‘J’ t t T 3 4 5 sound — ‘J’ did not appear in the Latin alphabet w w W 6 7 until about the mid-1500’s. x x X 6 Later Semitic scripts z 9 This section includes scripts that were invented after 1000 bc and used in the Middle East. Table 10 shows the evolution of the modern He- brew and Arabic7 scripts. 5.4 Cypriot The Cypriot script was a syllabary used in 6.1 during the approximate period between 1000 and It is believed that the script 200 for writing Greek. It has a relationship bc was invented on the order of the Persian king Dar- to Linear B as it includes some of the same signs. ius I for use on royal monuments. The script was Towards the end of its life few people could read only in use between about 500 and 350 bc. the script, so inscriptions were written using both Old Persian was a syllabary with 36 glyphs. the syllabary and the Greek alphabetic characters. There were also 5 ideographs, some with multiple These bilinguals made it relatively easy to decipher forms, for the words king, country, earth, god and the script, a task that was essentially completed by Ahuramazda (the Persian god), together with a word the last quarter of the nineteenth century. divider. Numerals were also represented. Like Linear B, the Cypriot script has no rela- Somewhat surprisingly the decipherment of Old tionship with the Greek abecedary apart from the Persian led directly to the decipherment of the far fact that both can be used for writing the same lan- older Sumerian and Babylonian cuneiform scripts. guage. The basic work was done by Georg Friedrich Grote- Table 8 shows the Cypriot syllabary. fend (1775–1853), a high school teacher in G¨ottingen 5.5 Runic and was completed by Henry Rawlinson [Adk04]. It was generally assumed that because of the limited The runic alphabet, which is not shown in Table 7, is number of signs the script was alphabetic, the slant- known as futhark after its initial characters. It was ing wedge was probably a word divider and it was used, with local variations, in the Germanic, Scandi- written left to right. navian and Anglo-Saxon countries until shortly after Grotefend started with two texts which were in- printing was invented. Scholars are unclear as to the scribed above doorways in the ruined city of Perse- origins of the futhark abecedary, but there are ob- polis. The first was: vious correspondences between some of the glyphs and the Phoenician and Etruscan ones, while others daryvuS:xSayoiy: have no resemblance at all. vzrk:xSaoiy:xSa Like the Phoenician alphabet, the names of the 7 futhark characters have meanings. The ordering of Arabic actually has 28 characters and many more glyphs as they change depending on the position of the character in a the characters, together with their names and mean- word, but I have only shown the characters that were derived ings, is shown in Table 9. from the Phoenician script. 212 TUGboat, Volume 26 (2005), No. 3

Table 9: The Futhark abecedary Glyph Name Meaning Glyph Name Meaning F feof, feh, fe wealth Y hic, ih, eoh U ur, hur auroch P peord ˆ thorn X eohlx A æsc, os oak tree S sigel sun R rad, rat riding T tir name of a star? K ce, kaun torch B berc, birth birch tree G gebu, gyfu gift E hæc, ech, eh horse W wen joy M man man H hegl, hagal hail L lagu water or sea N nyd, nod need or hardship ˘ I is ice D dag, dæg day J ger, yr, ar year O o, oe mouth : punctuation

Table 10: Evolution of Middle Eastern scripts Name Hiero. Proto. Phoen. Aram. Nab. Hebrew Arabic aleph ’A ’ ’ ’ ` X @ beth bB b b b a j H. gimel g g g g b / h. daleth dD d d d c F X he eE h h h d Y è vav w f, w w w e u ð zayin z z z z f P heth Hh H H H g x p teth T T T T h H yod yY y y y i d ø kaph kK k k k k P ¼ lamed lL l l l l ? È mem m m m m n n Ð nun n n n n p D à samekh s s s q € ayin ‘O ‘ ‘ ‘ r e ¨ pe ? pP p p p t b H sade xX x x x v  qoph qQ q q q w † resh rR r r r x Q P shin S S S S y E € tav t t t t z + H TUGboat, Volume 26 (2005), No. 3 213

yoiynam:xSayoiy: dhyunam:ViStasphy Table 11: The Old Persian syllabary a:puC:hxamniSiy:h a i u a i u y:imm:tcrm: 13 a i u aKunuS k k K f f And the second was: x x b b xSyarSa:xSayoiy: g g G m m w M vzr c c y y k:xSayoiy:xSayoiya j j J r r R nam:daryvhuS: t t T l l xSayoi th o v v V yhya:puC:hxamniSiy: ¸ca C s s d P D S At the time of the (559– d ˇsa n n N z z 331 bc) in Persia the Zoroastrians were prominent p h together with their sacred text, the Zend-. p h This had recently become known in as well as some Pehlevi8 texts translated and published by Silvestre de Sacy in 1793 in his M´emoires sur di- tain near Behistun in western . Henry Rawl- verses Antiquit´esde Perse. Grotefend knew of a inson (1810–1895), who was serving as the British formula from these texts that went ‘X, king . . . , king military advisor to the brother of the Shah of Iran, of kings, son of Y. . . ’. He guessed that the Perse- began to copy the inscriptions in 1835. Among other polis inscriptions might embody the same formula things, this involved dangling from to get at and that the second sign group in each inscription some of the texts as they were on a cliff more than (xSayoiy) might be the word for king, 300 feet up. In all, it took him ten years to copy and the later repetitions equate to ‘king of kings’. 414 lines. It turned out that the inscriptions were In the first inscription the first group trilinguals, in Old Persian, Elamite and Babylonian. daryvuS Rawlinson effectively completed the decipherment of should be the name of a king, say A for the sake Old Persian which then enabled him to tackle the of argument, and similarly in the second inscription Babylonian, which became the key to other cunei- the first group form scripts. xSyarSa Table 11 shows the Old Persian syllabary. would be the name of another king, say B. Grotefend Some of the other glyphs include: X (king), noted that A also occurred in B’s inscription, so q (country), L (earth), B (god), and F (Ahu- that the second one might read ‘B, king. . . , king of ramazda). kings, son of king A. . . ’. He guessed again that the Examples of numerals include: 1 (1), 2 (2), 3 inscriptions might have something to do with Darius (10), 4 (20), 5 (100). and Xerxes, and in which case A would be Darius As a final note, the transliteration of the second and B his son Xerxes. of the two inscriptions that Grotefend tackled is: Using forms of Darius and Xerxes which he de- xa-ˇsa-ya-a-ra-ˇsa-a-:xa-ˇsa-a-ya-tha-i-ya-: --ra- rived from an amalgamation of , Greek and ka-: xa-ˇsa-a-ya-tha-i-ya-: xa-ˇsa-a-ya-tha-i-ya-a- Hebrew he suggested that the signs in the two names na-a-ma-: da-a-ra-ya-va-ha-u-ˇsa-:xa-ˇsa-a-ya-tha-i- should be read as d/a/r/h/e/u/sh for Darius and ya-ha-ya-a-: pa-u-¸ca-: ha-xa-a-ma-na-i-ˇsa-i-ya-: kh/sh/h/e/r/sh/e for Xerxes. With these values the signs for king would read kh/sh/e/h/?/?/h. In the and which, when translated, means: Avesta he found the kingly title khscheio and he took Xerxes, the great king, the king of kings, the son of this as confirmation that he was on the right track Darius the king, an Achaemenian. and the language of the inscriptions was Avestan. He was correct. 6.2 Aramaic With the limited number of inscriptions avail- The Aramaic script is an early offshoot from the able, work could not proceed further. However, huge Phoenician and was used between about the tenth inscriptions were discovered on the side of a moun- and second centuries bc in the Middle East. The 8 Pehlevi (or Pahlevi or Pahlavi) was a Persian dialect of Aramaic script also branched and led to both mod- the Sassinide period (3rd–7th century ad). ern Arabic and Square Hebrew scripts. 214 TUGboat, Volume 26 (2005), No. 3

The script is alphabetical and consists of 22 [Cha58] John Chadwick. The Decipherment of consonants. Linear B. Cambridge University Press, abgdhwzHTyklmnsopxqrSt 1958. [Cha87] John Chadwick. Linear B and Related 6.3 Nabatean Scripts. Reading the Past. University of The Nabatean script is an offshoot of the Aramaic California Press, 1987. script and was in use in an area centered around Pe- [CM98] Mark Collier and Bill Manley. How To tra — the ‘rose-red city half as old as time’ — roughly Read Egyptian Hieroglyphs. University of during the period between the fourth century bc and California Press, 1998. the fourth century ad. It is a direct ancestor of the [Coe99] Michael D. Coe. Breaking the Code. modern . Thames & Hudson, 1999. Like other Semitic scripts it is alphabetical and [Dav87] W. F. Davies. Egyptian Hieroglyphs. consists of 22 consonants. Reading the Past. University of California abgdhwzHTyklmnsopxqrSt Press, 1987. [Fri89] Johannes Friedrich. Extinct Languages. 7 Remarks Dorset Press, 1989. (Translated from the original German Entzifferung The result from formula 1 is certainly an approxi- Verschollener Schriften und Sprachen by mation. I applied it to the two Old Persian texts in Frank Gaynor). section 6.1 which together contain 150 signs with 22 [Gor87] Cyrus H. Gordon. Forgotten different kinds. The estimated number of signs is Scripts: Their Ongoing Discovery S = 1502/(150 − 22) − 150 = 25.78 and Decipherment. Dorset Press, 1987. [Hea90] John F. Healey. The Early Alphabet. which is somewhat under the actual value of 36 for Reading the Past. University of California the syllabary. The much shorter made up text on Press, 1990. page 202 consisting of 70 characters of 29 different [Ifr00] Georges Ifrah. The History kinds gives of Numbers. John Wiley & Sons, S = 702/(70 − 29) − 70 = 49.5 2000. (Originally published as Histoire universelle des chiffres. Robert Laffort, which is a significant overestimate. However, com- , 1994.). bining the three texts gives 220 total characters with [Mal91] J. P. Mallory. In Search of the 33 different kinds, resulting in Indo-Europeans: Language, S = 2202/(220 − 33) − 220 = 38.8 and Myth. Thames and Hudson, 1991. [Nak80] Akira Nakanashi. Writing Systems which is close to the actual number. of the World: Alphabets, , The listed below are among the more ac- Pictograms. Charles E. Tuttle Company, cessible sources describing the development of the 1980. alphabet and the , and of decipherments of archaic scripts. [Pop99] Maurice Pope. The Story of The fonts used in this article can be obtained Decipherment: From Egyptian Hieroglyphs to . Thames & Hudson, 1999. from CTAN (the Comprehensive TEX Archive Net- work). The Arabic script came from Klaus Lagally’s [Rob95] Andrew Robinson. The Story of Writing: arabtex package in the languages area, and sim- Alphabets, Hieroglyphs & Pictograms. ilarly the Hebrew script is from hebrew/hebtex in Thames & Hudson, 1995. the same area. The Trajan font is in the fonts/ [Rob02] Andrew Robinson. Lost Languages: The trajan directory. All the other scripts are in the Enigma of The World’s Undeciphered fonts/archaic directory. Scripts. McGraw-Hill, 2002. [Wal87] C. B. F. Walker. Cuneiform. Reading the References Past. University of California Press, 1987. [Adk04] Lesley Adkins. Empires of the Plain:  Peter Wilson Henry Rawlinson and the Lost Languages 18912 8th . SW of Babylon. Harper Perennial, 2004. Normandy Park, WA 98166 [Bud89] E. A. Wallis Budge. The Rosetta Stone. USA Dover, 1989. herries.press (at) earthlink.net

Top View