Wednesday, March 19, 2014

Hebrew syllabary

The Lost Link
The Alphabet in the Hands of the Early Israelites

Brian E. Colless
Massey University, Palmerston North, New Zealand
(Go there to see the pictures)

I have made a surprising discovery about the way the alphabet was used by the Israelites in their early period of settlement in the Promised Land, that is, the time of the Judges (including Samuel) or, archaeologically speaking, Iron Age I (c. 1200 to 1000 BCE).

I feel like the woman in the parable (Luke 15:8-9) who lit a lamp, poured light on the scene, and searched diligently for a lost coin, one of her ten silver drachmas; when she found it she rejoiced and rushed to tell her friends and neighbours. So here is my ten drachmas’ worth to share with everybody, as I am confident that I have alighted on something significant that has been overlooked for three millennia. This lost object, now ready for our inspection, is just one of a whole bunch of keys that can open doors to our understanding of the early alphabet, with regard to its features and functions. Indeed, it is a missing link in a ten-point theory of the origin and evolution of the alphabet.

The main difficulty we encounter, as practitioners of ancient Hebrew epigraphy, is the dearth of documents available for study, from those early centuries of the Iron Age. The few inscriptions we do have are brief and frequently fragmentary.[1] However, things are now improving, and we possess two five-line texts, namely the Izbet Sartah Ostracon[2] and the Khirbet Qeiyafa Ostracon[3]. 

The Qeiyafa Ostracon came to light in Khirbet Qeiyafa, a fort overlooking the Valley of Elah, where David confronted Goliath (1 Samuel 17); the site has been plausibly identified as Sha`arayim (mentioned in 1 Samuel 17:52), but other names have been proposed,.

A tentative drawing of the Qeiyafa Ostracon (BEC)

On this potsherd from three thousand years ago there are five faded lines of alphabetic letters, written in ink; and some of the signs have their upper part missing, because of damage to the top of the ostracon. The text runs from left to right (the opposite of later Hebrew writing). The first letter on the first line (top left corner) is ’Aleph, in a reclining attitude. The better-known upright stance (Greek Alpha and Roman A) appears at the start of lines 4 and 5. The original form, the head of an ox with horns, is found near the end of the top line. Closer examination reveals that other letters likewise have variations in their shape and stance. Notice that the letter Shin stands at the beginning of the second line in a figure-3 pose, but further along the line it looks like W. The numerous instances of Mem (all vertical, never like the horizontal Greco-Roman M) have different numbers of curves (or water waves).

The question that begs to be asked (though nobody has dared to articulate it) is whether all the variations are arbitrary (just vagaries of the writer’s style of handwriting, like the three different renderings of P in my personal repertoire) or whether the differences are deliberate and meaningful.

The common assumption is that the early Hebrew alphabet was like the Phoenician alphabet of the Iron Age: it had twenty-two letters and each of them stood for a particular consonant, and of course there was no indication of the vowels in writing. The dots and dashes found in Hebrew Bibles for indicating vowels were much later inventions of scribes who were concerned that the Sacred Scriptures should be recited correctly. Eventually, later in the Iron Age, vowel markers were inserted in a Hebrew text, using Waw for showing the presence of a long u or o, Yod for long i or e, and He for long a; these are known technically as matres lectionis (mothers of reading, offering sustenance to the reader, so to speak). This idea was not used in Phoenician inscriptions, but it is still employed in Modern Hebrew writing.

Returning to our question regarding the possible significance of the variations in the letters on the Qeiyafa Ostracon, we may now ask whether the various forms and stances of each letter constituted a way of denoting particular vowels. Actually, there was already an analogy for this in the West Semitic cuneiform alphabet, which was particularly associated with the city of Ugarit (on the Syrian coast, east of Cyprus) but it is also attested throughout the Levant. The scribes of Ugarit, in the Late Bronze Age II, immediately preceding the Iron Age, had a cuneiform sign for each consonant, made up of wedge-shaped marks, and I am convinced (having demonstrated this elsewhere) that each was constructed with the image of the corresponding letter of the Semitic alphabet in mind.[4]  This is the relevant detail: on the clay tablets of Ugarit there were three distinct ’Aleph characters, representing ’a, ’i, ’u.

This phenomenon could account for the trio of ’Aleph forms on the Qeiyafa Ostracon. But my idea goes further than that: in the early “Israelian” Hebrew alphabet each of its twenty-two letters had two additional forms, making a total of sixty-six signs. In fact, the system was not a simple consonantary (which the Phoenician alphabet certainly was, in the Iron Age) but a syllabary. Incidentally, there has long been a discussion about the nature of the Phoenician alphabet, and whether it should be understood as a syllabary, with each sign standing for a consonant with any vowel; but that is a different matter. What I am proposing here is a syllabic alphabet.

How can this syllabary hypothesis be tested? A first step would be to compare the letters found on the Qeiyafa inscription with their counterparts in the Izbet Sartah text.

Tentative drawing of the Izbet Sartah Ostracon (BEC)
The boon of this document is the copy of the scribe’s own script in the fifth line, running from the left (sinistrodextrally) with the familiar Aleph, Beth, Gimel, Daleth, which would become Alpha, Beta, Gamma, Delta when the Hellenic alphabet was constructed from the West Semitic consonantary, and when Aleph (the glottal stop) became the sign for the vowel a, with no consonantal reference. The copy of the alphabet presented to us on the Izbet Sartah Ostracon has the expected twenty-two letters, though some of them are hiding from us, and several of them are not in the order we know from the later Hebrew and Greek alphabets.

Habitually in this field of endeavour, the assumption is made that when the letters in this abecedary (or “abagadary”) are used in a text, the reader will have to supply the appropriate vowel in every case. However, in the four lines of writing above the Izbet Sartah alphabet (which are an interesting personal message from the writer himself, in my interpretation of them) we are offered Aleph in more than one stance (though the inverted A-form, the ox-head, is absent here). And when we have established the shape of the Lamed in the bottom line (more like a 6 than a 9), we look higher and see 9s as well as 6s (with the added complication that this scribe makes it difficult to distinguish B and L).

Consider also Kaph: we can see the origins of Greek Kappa in the model provided in line 5, but at the beginning of line 2 it is quite different, being a bisected right angle on a stem; I suggest (after experimenting with possible words and meanings in both texts) that the K-form says ka, and the trident is ki. My detailed observation of the letters in this line, in comparison with examples on the Qeiyafa inscription and other related documents, leads me to suspect that this inventory is indeed an abagadary, in that every sign has the vowel a built into it: ’A, BA, GA, DA, and so on to TA. The Taw here is not a simple + cross (like a plus sign) since the cross-beam points north-east; and if we focus on a trio of them, one above another, in lines 1, 2, and 4, we see the middle one has its cross-bar pointing north-west, and I have reasons for believing it is TI.

Now, if we take the Izbet Sartah TA to the Qeiyafa text, this matches the third letter in the top line, and the presumed TI is the penultimate sign in the bottom line. The TU is lurking in line 3 (a small character) as a cross that is shaped almost like a multiplication sign. If we apply this information to Phoenician inscriptions from the same period we might expect to see the TA syllabogram functioning as Taw, or even a perfectly even cross (+); but the TI is predominant.

Looking at Shin, we know two forms, “W” (SHI) and “3” (SHA in the Izbet Sartah abagadary), and the Qubur Walayda Bowl (pictured below) provides the third, a reversed “3”; the SHI sign (W) is what we see in the Phoenician realm, as well as in later Hebrew inscriptions, notably the Gezer Calendar[5], which show that the Phoenician style was adopted, in the 10th century BCE, during the reigns of David and Solomon, when commercial and diplomatic relations were flourishing between Jerusalem and Tyre.

A supposition is rising up before our eyes: the Izbet Sartah abagadary (with its a-vowel syllabograms) is not the place to look for the originals of the letters in the later consonantal alphabet, because this simpler system was made up of i-syllable signs (but with the vowels ignored). Fortunately we have at our disposal a list of the consonantal letters from Iron Age II, on the Tel Zayit Stone, and it seems that this document confirms the supposition.[6]

Thus, the Zayit K is not the KA of the Izbet Sartah abagadary, but the KI at the start of line 2 (as mentioned earlier), which is a trident. The Qeiyafa KI (in the word MLK “king”) apparently has no stem, but this is likewise the case with all the Phoenician examples.

Seeking a Phoenician counterpart for one of the three Aleph characters, we are at a loss, but perhaps the key feature is not the stance but the crossbar, which breaks through one side or both sides. The Zayit Aleph is typical, pointing leftwards, the direction of the writing, and having a vertical line cutting through it. Here we must acknowledge that the various scribes had their idiosyncrasies (whether regional or personal) and possibly made their own rules for distinguishing the syllabic forms of each letter.

Samek in the international consonantal alphabet was a “telegraph pole”, a stem with three (or just two) crossbars; it is actually an Egyptian djed, representing a spinal column. However, this character does not appear in any of the Iron Age I inscriptions, because they have a fish for their Samek. That is what we see in the Izbet Sartah abagadary between Nun and Pe; on the Qeiyafa text, in line 4, between Y and D (notice the dots indicating a fan-tail in each case); also on the Beth-Shemesh Ostracon; but prejudgement (rather than blind prejudice) does not allow most epigraphists to see it, because they are convinced that the fish in the Bronze Age proto-alphabet was D (from dag “fish”). Possibly the tree-shaped Samek functioned as SI in the syllabic alphabet, and it has not turned up in the extant inscriptions. There were presumably 66 signs, and we certainly have not seen them all yet.

‘Ayin in the Phoenician alphabet, as also on the Zayit Stone, is a circle, but without the centre-dot that is characteristic of the sign in the south in Iron Age I. However, on the Beth-Shemesh Ostracon[7] there is one of each, leading to the suspicion that the empty circle is the syllable ‘I. The sign for ‘U remains to be identified.

Mem has a variety of forms in the documents, but there are three basic types: the Phoenician M is presumably MI and it has the top stroke pointing NE, and the bottom stroke SW (the number of intervening angles varies for MI in Israel); MA has the top as NE and the bottom as SE (and only one angle in between); MU has the top as NW and the bottom as SW (with two angles or only one in between). Confusion could occur between MA and SHU, and also between SHA and short versions of MU, but the Sh-signs have horizontal tops and bottoms.

Looking back to the beginnings, the original prototype of the alphabet came out of the West Semitic syllabary in the Bronze Age.[8] At least eighteen (18/22) of the letters that survived into the Phoenician and Hebrew consonantal alphabet of the Iron Age were already in the syllabary; this can be seen on the table of alphabet evolution appended below (note the BS [Byblos script] column in the Phoenicia section). For the most part, the syllabograms with –a were chosen for this purpose. The proto-alphabet was patterned after the Egyptian writing system, which did not indicate vowels, but the characters could function as logograms, ideograms, and rebograms (rebuses).[9] We see this in the proto-alphabetic inscription from Wadi el-Hol in the Egyptian desert, which speaks of feasting in celebration of the goddess ‘Anat, in agreement with the Egyptian inscriptions reporting similar carousing in honour of Hathor.[10]

The new Iron Age syllabary did not necessarily select the direct descendant of these originals for the a-set:  for example, the Aleph corresponding to the ox-head is not ’a, but little more can be said on this score at present.

Suddenly we realise that we are hearing the vowels as well as the consonants! The writer of the Izbet Sartah says, in my reading of his text (line 1b-2a): “I see that the eye gives the breath of the sign into the ear”. In this statement he uses the ‘Ayin sign for “see” and then for “the eye”, continuing the practice that goes right back to the inception of the proto-alphabet, and also to the West Semitic syllabary that preceded it, whereby the pictophonograms could act not only as consonantograms or syllabograms, but also as logograms, ideograms, and rebograms (rebuses), as happened with the hieroglyphs in the Egyptian writing system, which influenced the formation of these two West Semitic scripts.

Thus, the early Hebrew alphabet of Iron Age I was not the same as the Phoenician alphabet of the Iron Age: it had sixty-six letters (not simply twenty-two), and each of them stood for a particular open syllable (consonant plus vowel, not merely a single consonant), and so there was every indication of the vowels in writing. Moreover, the signs were multifunctional, a concept that is never countenanced in the standard manuals of West Semitic epigraphy.

In summary, here are some indicators for distinguishing the two early Iron Age alphabetic scripts, which we might call (1) the neo-syllabary and (2) the international consonantary.

(1) Various forms for each letter (’abugida syllabary)
(2) Single form for each letter (’bgd consonantary)

(1)  Dextrograde, sinistrodextral (L > R)
(2)  Sinistrograde, dextrosinistral (L < R)

(1)  Fish for Samek
(2)  Spinal column (djed) for Samek

(1)  Dot in circle of `ayin
(2)  No dot in circle of `ayin

(1)  Vertical and horizontal forms of Sh-sign (3 and W)
(2)  Horizontal Sh only (\/\/)

(1)  Logography and Rebography
(2) Consonantal writing only
In the light of these criteria, we may consider some examples.
The broken inscription on the Ophel Pithos from Jerusalem is highly problematic, not having the key indicators (Samek, ‘Ayin, Shin), and with uncertainty about the direction of the writing.[11] It appears to have two forms of Nun (NI and a reversed version) or possibly two different instances of Mem; its Het is unusual, lacking a central crossbar, and so it seems to be neo-syllabic, possibly H.U but not H.I; the Mem (first letter on the left) seems to fit the MI mould, rather than MU or MA, and so it is not prescriptive either way; the letter following could be R (facing in the wrong direction) or Q (which should have the stem piercing the oval for QI and Phoenician Q), and the next one P (a reversal of the Phoenician P) or L (inverted and reversed). It is thus anomalous as a Phoenician-style text, and this perhaps supports it as neo-syllabic; and so it might have been written before David established his capital there. But it was inscribed before it was baked, and it was not necessarily made in Jerusalem.
Fragmentary pithos inscription from Jerusalem
Then there is the Qubur el-Walaydah bowl, usually dated early in the Iron Age, and is quite instructive as far as it goes.
Drawing of the QW bowl (BEC)
Transcribing the writing (from left to right) with consonants only, I read:
                  Sh M B? ` L | ’ Y ’ L | M H.
The bulk of the text is a personal name, the second part being the patronymic. As stated above, the Sh-sign (not “3” SHA nor “W” SHI) would be SHU.  The Mem should be MI (though it has more angles than the Phoenician form). Thirdly B or P (but BA seems more likely than PA, because it is the god Ba`al). Then comes ‘Ayin with no dot, so it is ‘I. The Lamed here is not the same as the one further on, and I will not make a decision on their respective sounds; but it might be LI. Hence Shumiba`ili  (“name of Ba`al”). Regarding the Mem Het combination (apparently separated from the rest by another dividing stroke), this could say MAH.U “a fatling” (a sacrifice, as at the end of the Wadi el-Hol horizontal inscription).

Another intriguing inscription comes from Gath, the hometown of Goliath.[12]
Gath Ostracon
The line of writing probably runs from right to left, and this suggests it is the international alphabet, not the neo-syllabary; but the Aleph has the wrong stance (it is upright, not reclining) though it has the characteristic extension of its crossbar. The two instances of L seem to be the same, and they are reversed versions of those on the QW bowl above, and also of the Lamed in the Izbet Sartah abagadary. It is hard to determine whether the two Y-shaped characters represent Waw or Yod. The first Taw might be meant to be TU or TA, and the second one TI, or else the stance has no particular significance. It is supposed to say ’LWT WLT (the dividing line, as on the QW Bowl, could signify “son of”), and a resemblance to GLYT (Goliath) is pointed out.
The Gath inscription is certainly Philistian, and perhaps its script is a local version of the consonantary or the syllabary. The Qeiyafa Ostracon is from a fort in the vicinity of Gath, but I have reasons for believing its script and language are Israelian. The Zayit alphabet is clearly the international consonantary, and this site is also in Philistia, near Gaza, but it could be Israelian, in the kingdom of David and Solomon. The Qubur Walayda inscription is supposed to be early, and its geographical position (south of Gaza in the Negev) hints that it might be Philistian rather than Israelian; the name Shumi-Ba`ili is not decisive.
Consequently it may well be that the new syllabary was not only used by Israelites but also by Philistines, and possibly Jebusites (in Jerusalem, on the Ophel pithos). In any case, my guess is that the change from local syllabary to international consonantary came about when the Davidic dynasty established cultural and commercial ties with Phoenicia (2 Samuel 5:11-12, 1 Kings 5:1-18, Hiram of Tyre) and adopted the international consonantal alphabet, as used in Tyre, Sidon, and Byblos (the chief supplier of our inscriptional evidence).
And now a question about priority (or priorness): which came first? Was the Phoenician consonantary a reduction of the Israelian neo-syllabary, employing only the -i characters? Or was the Hebrew neo-syllabary an expansion of the Phoenician alphabet, which became the i-column in the new construction?
There are no known alphabetic inscriptions from Phoenicia in the Bronze Age. It is not credible to surmise that the Phoenicians were not writing and recording in that era, but we might assume that they were favouring the original Byblos syllabary, which is attested all around the Mediterranean region and beyond (notably Jamaica[13] and Scandinavia[14]); and their written records would have been on perishable papyrus.
Nevertheless, four Phoenician alphabetic inscriptions (or graffiti) have turned up on clay tablets, in Mesopotamia (from the Sealand, at the Persian Gulf) from the end of the First Dynasty of Babylon, that is, around 1600 BCE. Amazingly the letters have the same forms as the Iron Age consonantary. One of them is a transcription of the name Ali-dîn-ili. For our purposes it is unfortunate to see so many –i syllables in it, but we can say that the same form of Aleph is used in Ali and Ili (an acute angle with a vertical line running through both sides, and pointing leftwards, the expected direction of the writing). Another graffito begins with that form of Aleph, and, judging only from drawings, both texts have the regular forms known from the Iron Age (the Tel Zayit alphabet is invoked by the reporting researcher, for comparison). One exception is Yod, which is uncharacteristically reversed. This surprising evidence needs closer scrutiny and confirmation of its date.[15]
In the nineteenth century the bones of the Ichthyosaurus lay “unpublished” in glass cases in the Ashmolean Museum for a hundred years or more; they were disturbing to the consensus. In our own times, photographs of two ’abagadaries of the proto-alphabet (which were discovered in southern Egypt and included by William Flinders Petrie, in 1912, in the frontispiece of his book The Formation of the Alphabet) have been disregarded and not taken into account in all the published discussions and speculations that have gone on in this realm; and they are not mentioned in any of the handbooks on the subject.[16]
The climate is changing drastically around us, and apparently the paradigm has now shifted in West Semitic epigraphy. Still, this neo-syllabary (which has been posited here) could be a hypothesis that can be easily falsified; but the variation in forms of letters in a single inscription such as the Qeiyafa Ostracon requires an investigation.
I have chosen to announce my findings here, because of the strong connection with ASOR that the forerunners had, namely William Foxwell Albright and Frank Moore Cross, and I regret that they did not live to share in this new knowledge. Unfortunately their successors sometimes fail to test the theories of these masters, upholding the proposed identifications for the signs of the proto-alphabet without taking sufficient account of other possibilities.
I have been inspired by the words of another keen practitioner of this art and science, namely Gordon J. Hamilton, who said (on the internet) that the recent discoveries of such inscriptions should have the effect of “motivating researchers to dig deeper and reconsider previous views”.
We must also try harder to read these important texts. In line 3 of the Qeiyafa Ostracon, I discern this sequence: M T G L Y T B ` L D W D, and the syllables seem to reveal GULIYUTU and DAWIDU, and the meaning is: “Goliath is dead, David has prevailed (ba`ala)”.[17]
Ten-point theory of the origin and evolution of the alphabet.[18]
 (1) Before 3000 BCE in Mesopotamia a complex pictophonographic logosyllabary emerged; the rebus principle produced numerous phonetic syllabograms from its logograms; its pictorial and symbolic characters turned into cuneiform script, using wedge-shaped components. Five vowels were represented (i e a o u).
(2) Before 3000 BCE in Egypt a complex pictophonographic logoconsonantary was created; its pictorial and symbolic hieroglyphs became stylized and unrecognizable (in the Hieratic script), but the original shapes of its logograms and phonograms (simple and complex consonantograms) were preserved for monumental inscriptions. Generally, vowels were not represented.
(3) Before 2300 (in the Early Bronze Age) in Canaan (Syria-Palestine), probably at Gubla (Byblos), a simpler pictophonographic logosyllabary was constructed, largely from Egyptian characters, and using the acrophonic principle (a modification of the rebus principle) to form syllabograms (representing a single consonant with a vowel), which could also function as logograms and rebograms. Twenty-two consonants and three vowels (i a u) were represented.
(4) This became the model for other acrophonic syllabic scripts: Aegean (Cretan and Cyprian, 5 vowels) and Anatolian (Luwian, 3 vowels); also American (notably the Maya acrophonic logosyllabary, with 5 vowels).
(5) After 2000 BCE (in the Middle Bronze Age) in Egypt, apparently, the West Semitic proto-alphabet was constructed as a simple logoconsonantary, largely from elements in the West Semitic syllabary and thus based on Egyptian hieroglyphs; its phonograms were acrophonic consonantograms, which could also function as logograms and rebograms. The number of consonants was expanded from 22 to 27 (or possibly more). No vowels were represented.
(6) This proto-alphabet spread to the Levant (in the Hyksos period, 17th century BCE?) and was used alongside the syllabary (in Egypt and Canaan). No vowels were represented. The number of consonants was retained at 27, but this was, sooner or later, reduced to 22 phonemes.
(7) In the Late Bronze Age, apparently at Ugarit, a cuneiform consonantary was devised, with cuneiform characters based on the pictorial forms of the proto-alphabet. There was a long version, with the full complement of consonants, and syllabic signs for ’a ’i ’u; and also a short version used beyond Ugarit.
(8) Early in the Iron Age, in Israel (and Philistia?) a syllabic alphabet appears, “the neo-syllabary”, a new development employing the stylized forms of the proto-alphabet, with a reduced number of consonants (22, the same as in the earlier West Semitic logosyllabary), and with three vowels (u a i) represented by changing the shapes and stances of the letters. The practice of logography persisted, and the signs were multifunctional, as in the proto-alphabet.
(9) In the Levant, in the Iron Age, a simple consonantary with 22 letters, conventionally known as the Phoenician alphabet, became the international writing system for West Semitic languages (Phoenician, Hebrew, Aramaic, Moabite, and others). In some cases matres lectionis were added, to compensate for the lack of vowel signs.
(10) In the Hellenic regions an alphabet was invented, with the elements of the West Semitic consonantary put to new uses, including the innovation of vowel-letters (“vocalograms”).
Table of the evolution of the Alphabet
Note that the neo-syllabary of Iron Age I is not shown

[1] Early Iron Age inscriptions, with comparative tables of signs:
Fig 8: Tel Zayit abagadary
[3] Qeiyafa Ostracon: Photographs and drawings of the inscription appear on the Qeiyafa website:
[6] Tel Zayit ’bgdary (see also note 1 above):
[9] Technical terms (some are neologisms) in “paleogrammatology”:
[12] Gath Ostracon:
Maeir, A.M., Wimmer, S.J., Zukerman, A. et Demsky, A. 2008. An Iron Age I/IIA Archaic Alphabetic Inscription from Tell es-Safi/Gath: Paleography, Dating, and Historical-Cultural Significance. Bulletin of the American Schools of Oriental Research
[15] Four Phoenician graffiti on Babylonian clay tablets:
Laurent COLONNA D’ISTRIA Université de Genève, Épigraphes alphabétiques du pays de la Mer, N.A.B.U. 2012, 3 (septembre), 61-63
Dalley S., 2009, Babylonian Tablets from the First Sealand Dynasty in the Schøyen Collection, CUSAS 9
[17] My work-in-progress on the Qeiyafa Ostracon may be viewed here:

No comments: