Addressing issues of origin in archaic alphabetic writing
A multitude of questions and corresponding number of theories have been offered on the origin of our writing system, the alphabet. We have considered in detail the question of the systematic relationship of archaic alphabetic writing to other Mediterranean scripts in the context of Bronze age elsewhere [1]. Briefly, these results show that several ancient scripts namely Linear A, B and C (a term used for the Cypriotic syllabary [2]), Levantine Protolinear (related Levantine writing before 1050 BC), Phoenician abjad (after 1050 BC, the conventional date for delineation [3]) and Greek alphabet have related morphologies of corresponding symbols that produce a unique pattern of systematic relationships.

Remarkably, among the countless theories, often based on preconceptions of political ideology, ethnicity, religion or even ‘racial superiority’ [4], there has been little attention paid on the characteristics of the alphabet qua organomenon, that is, as an organized system. Following our analyses of systematic relationships, we revisit here elements of the internal structure of the system, with a particular focus on vowels, in order to address the issues of origin, whether the system was an invention and the context of its genesis.

The vowel harmonic
It is generally agreed that the 23-letter script (A to Y) of Crete and the Doric islands is the earliest alphabet in the Aegean (Fig. 1) and indeed earliest full alphabet encoding both vowels and consonants [2, 5, 6]. Focusing on the internal structure of that script rather than the later Ionic alphabet that has become the basis for the more commonly known Greek alphabetic writing, it immediately becomes apparent that there are five vowels, A, E, I, O and Y, separated by arithmetically increasing intervals of consonants (3, 4, 5 and 6).This simple distribution creates a harmonic that might have served as a mnemonic device in the early days of transition between the syllabaries and the alphabet. Since we are treating the issue of the origin of the alphabet and its systematic relations to other scripts elsewhere, we will limit this communication to notes on the vowel harmonic consisting of these five symbols.

Aegean origin of vowel encoding: morphology and meaningful sequence
Basic structural elements of the early alphabet provide several clues which could help elucidate the issues of origin, place, time of the invention of the alphabet etc. As discussed separately [1], the five vowels of the Linear C rather than those of Linear B are used in the early alphabet arranged in a sequence that would have been perfectly meaningful (AEI OY) at least to the speakers of the dialects written in Linear B and C. On the other hand, much of the structure of the alphabet is closely related to one of the Ugaritic cuneiform scripts. The combination of these, along with other evidence, point to the direction of an Eastern Mediterranean origin resulting from a major interaction between cultures [1]. Apart from the results of systematic analysis of morphology discussed elsewhere, it makes perfect sense that the vowels would have been a contribution of the Aegean writing systems: Aegean and Cypriot syllabaries that have been deciphered (Linear B and Linear C) encode vowels, unlike most other scripts of the Bronze Age.

In Fig. 2 we reproduce part of the corresponding outlines of related symbol morphologies from [1]. Attention is drawn to the fact that there is no generally accepted decipherment of the Vinca or Linear A and hence correspondences are not based on homology but follow studies and suggestions by other scholars [7-10]. Further connections of the Linear syllabaries to the Vinca semiotic system have been proposed [8, 10]but can neither be confirmed nor denied on the basis of current evidence and the lack of a convincing decipherment. For the remainder five scripts, Linear B, Linear C, Levantine Protolinear, Phoenician and archaic alphabets, homologies are easy to find or (in the case of the Levantine scripts where vowels are not noted) deduce (Fig. 2).

The alphabet as an invention: the ‘signature’ in the nested order of vowel values
Morphology may be criticized, at least to a certain degree, as subjective; similarly the meaning of the vowel harmonic could be criticized as coincidental. There are other elements of the harmonic however that can be independently verified; one is the arithmetic distribution discussed above. Another is provided by the phonetic value of the related syllabograms and their relation to values and names of the classical alphabetic symbols. The vowel harmonic shapes are:

A   E   I   O   Y 				(i)

While the related syllabograms are 

AB38, B43, B36, AB61, AB10 		(ii)

Respectively (numbers refer to Linear B and A related symbols). For the benefit of simplicity we will accept here B36, (Ventris value [Jo] or [Io] [7, 9]) as more archaic in alphabetic writing than the straight-lined I (Ventris value [I]), while in archaic alphabets both forms were present and indeed the latter survived to this day (for full discussion see [1]). The Ventris phonetic values of the related Linear B symbols (Fig. 2) are:

E Ai Jo O U					(iii)

where the modern letter u expresses for simplicity a sound close to French /u/ and german /y/, which was encoded by archaic Y; the shapes Y and õ have been equivalent in Greek alphabetic writing. At first sight, there is a striking reversal of values compared to our own for the first two. However, this reversal is already present in Linear C where the related symbols have the values:

A E Jo O U					(iv)

Let us now compare the classical names of the vowel harmonic:

Alpha, Ei, Iota, Ou, U				(v)

Isolating the suffixes (v) becomes:

A-lpha, Ei, Io-ta, Ou, U			(vi)

Two things are immediately noteworthy in (vi). First, there is a remarkable conservation of the related syllabogram value as accepted today and presented in (iii) and (iv) in the surviving classical letter names. Second, the latter half of the corresponding syllabogram value consists of the first half of the next member of the harmonic for the four latter members, even if we accept the currently used decipherments as a guide. There can be little doubt that this order could not have been the result of randomness but was the conscious result of invention by someone who was familiar with the syllabic values of the Linear syllabary symbols.

What were the values of AB38 and B43?
If only the order of values was reversed in AB38 and B43, the correspondence of series (iii) and (iv) would have been perfect and the relationship between syllabaries and the alphabet established long ago. Sad though it may sound, history does not always progress in perfection. In fact it is tempting to theorise on the possible perception of the symbol values on the ears of the inventor based on the problems posed by the reversal in value of the two symbols between (iii) and (iv) and indeed the present phonetic system in Greek and most other alphabetic languages.

Let us first address the issue of the second syllbogram B43. This has the Ventris Linear B value [Ai] but its related Linear C morph has the unambiguous value [E]. That transition AI>E, in fact is a well known phenomenon that has occurred repeatedly in different contexts, not least in Greek dialect and modern state Greek, were what is written as –AI is pronounced as /å/, exactly the same as the value of symbol Å. Now, the story becomes far simpler if one supposes that the pronunciation of B43 has been in fact closer to how the very combination AI is pronounced, for example, in the context of today’s English (/å:é/). This would fit with the interchange- ability of the symbol for respective sounds in Linear B and C (see series (iii) and (iv)) and the classical Greek name for the symbol (Ei). Hence we may suppose with some confidence that for the inventor(s) and first users the value of the symbol used as a form source for E was:

Value{E}~ Ei   (vii)

Which leads to the question: what was the value of AB38, with current Ventris value [E] in the time of the alphabet invention? If contemporary Greek dialects (rather than the homogenised southern, official Greek) or modern languages in general are any guide the answer can only be: it probably varied. Bivalence or even polyvalence of symbol phonetic value, with a single letter representing different sounds in different contexts is, of course, well known. In a remarkable coincidence of inversal of values, the very name of the symbol A in English is /å:i/, (that is, what a Greek would write as E:I), although its value in the word cat is what a Greek would understand as /á/ and write as A.

This interchangeability of a and e is neither new nor exclusively contemporary. For example, Attic Greek wrote THN (equivalent of TEEN) and Doric TAN. So the problem could find an easy solution if the dialect of the inventor (or the one he was familiar with, if they were not a native speaker) wrote the symbol A for both *e makron and *a makron (both cases occur independently in later local Greek alphabets) or if he was aware of the two values [E] and [A] in Linear B and C and thought the two sounds close enough to be represented by the same symbol. On the other hand a slightly different hypothesis on the value of Linear symbol AB38 may provide a more parsimonious explanation.

A “long” value for AB38 and/or its ancestor: a hypothesis
As we have seen, the symbols related to A in Linear syllabaries have values [E] and [A]. Based on what is currently accepted, Linear syllabaries did not note length of vowels, while the exact phonetic values are, naturally, for the most part irretrievable. A further hypothesis can be posited based on the fact that what is written as A in classical time Doric presumably represents dialectal a makron but also the corresponding in other dialects e makron (sometimes written as a double vowel –EE in later Greek). These are descendants of the same ancestral phoneme developing differently in different dialects.

It seems plausible to suggest that that sound may have been an intermediate double vowel closer to [Ae]. Several other linguistic elements, like the name for the sun (helios, but also haelios, aelios, elios or even alios[11]) indicate that an -ae sound has remained in use in the Aegean till quite late in the classical years. Hence, it seems reasonable that a symbol for that sound may have existed at some earlier stage. Since it is clear that A-related symbols in Fig. 2, Linear AB38 (current Ventris LB value [E]) and Linear C (current value [A]) are morphologically close, it is plausible to suggest that a symbol with value [Ae] would be included in this lineage, with general morphology of two lines posited in angle, crossed by a third line (variants of A).

If this was the value of the source symbol for A on the ears of the inventor or even if he recognized a bivalence for the symbol encoding both [A] and [E], as is indeed the case for symbols belonging to this lineage in Linear B and C then it is not irrational to assume that:

Value{A}~Ae (viii)

Now, if these relations, (values for the first member of the harmonic: [Ae] and second: [Ei]) were the case for the script that was the source of symbol form for the alphabet, or if it was the case in the ears of the inventor - at least at the time, place and scribal context of the invention a further interesting, and truly elegant, relation emerges. By replacing (vii) and (viii) in (i) along with the known values/names for the other syllabograms, the phonetic value order of (i) would be:

Ae Ei Io Ou U (ix)

It is not just that such a series as (ix) is a most elegant nested harmonic, with the latter part of each value perfectly representing the beginning of the value of the next symbol in its syllabic context. If elegance were a secondary concern of the inventor the practicality of such an arrangement can hardly be overestimated: such a phonetic value sequence would make a lot of sense.

Imagine that the inventor, the mythical Palamedes, faces the problem: how do I teach scribes familiar with Linear B and/or C the use of one symbol for one sound that is so much more efficient in the cuneiform (presumably the Ugaritic)? This harmonic on one hand includes and fixes all the five vowel sounds of the Eastern Greek syllabic system (at least as preserved in the Cypriotic Linear C; in bold capitals in (ix)) values that can re-create all vowel sounds in Greek; on the other it combines them with presumed Aegean/ Western Greek (close to Linear B derived system current at the time at any rate) values of the respective symbols.

This provides an easy to memorise nested order based on the syllabogram phonetic values making the new system compatible and easy to adopt by all Greek-writing scribes, whether they wrote in Linear B or Linear C. The mnemonic device would have probably quickly died out once the alphabet was broadly adopted and Linear syllabaries became extinct, since there was little need for it; but some of its elements survived as a fossil in the names of the symbols, in some cases to this day. It is worth noting that the exact value of AB38 does not affect the remainder of the vowel harmonic which exhibits nested values anyway.

Epilogue
It is interesting that so many theories on the origin of the alphabet have been offered based on preconceptions of ideological, religious or even ethnic nature and so little attention has been paid to the organization of the system itself. A view of the alphabet qua organomenon shows that much of these preconceptions and monogenic theories in general are misplaced. The alphabet appears to have been a complex invention emerging from cultural interaction and plexis. Based on present evidence, there seems to be little doubt that the vowels were an Aegean contribution (for a debate on other contributions see [1])..

Many questions still remain on the origin of our writing system that merit further investigation. Whatever the ultimate answers might be, it seems clear that it was an invention that took place in a particular historical and geographical context and that many of the currently held beliefs on the issue and the history of late Bronze Age need revisiting.

Figure 1
Structure of the 23-letter alphabet of Crete and the Doric islands. Several letter variants existed in local alphabets, deriving mostly from Linear B symbols. Only major allelomorphs are presented here. Vowels are noted below consonants; they are separated by arithmetically increasing numbers of the latter. Note that several letters were subsequently added at the end of the sequence in other local alphabets.

Figure 2
Symbol forms in the lineages resulting to our vowels A, E, I, O and Y and their inscription of origin [1]. For values of respective symbols in deciphered scripts see text. The Linear C symbol [O] is too derived and shows little similarity to the remainder scripts. Two different major allelomorphs existed for I, already in archaic alphabets, related to Linear syllabogram [Io] (which appears to be the most archaic) and [I]. The latter, straight-lined I, survived to this day in alphabets, the former in abjads. Note that relations to the undeciphered Vince semiotic system are conjectural since even its grammatological status remains controversial. Symbols belonging to that system and Linear A are included following suggestions by other scholars (see text).