Exploring the patterns on the (in)famous Phaistos Disc

Here I am again, back to blogging after a brief pause - the FIFA football (soccer) world championship can really make one attach to the TV screens - even if your own country's team is not playing...
For everyone's favour, I am returning to the Phaistos Disc. This is an all-time hot topic of fame-hungry amateur would-be decipherers and pseudo-linguists, so I promise to be as cautious and methodical as possible. We have to maintain a strictly scientific appproach even when half of the world love to think otherwise. Just to make clear: my objective here is not to "read" the disc word-by-word, but to understand a bit more of its structure, and thereby - so I hope - reveal something of its true meaning.


After a brief look at the disc, it becomes clear that the text recorded here is strikingly repetitive. Not only words are repeated, but entire phrases, even complete sentences. As problematic these repetitions in interpretation might be, they enable us to reconstruct the sentence-boundaries on the disc. The rules are quite simple: Whenever a complete phrase starts to repeat, it is at the start of a new sentence. Pertial repetitions may hint at parts of a compound sentence. With this method, we can get a structured version of the disc's contents (see the next figure below):


The disc actually contains two almost equally long texts. The texts occupy the shape of two snakes, and seem to be continuous from one side to the other. Though side A was historically considered to be the initial one by Evans, just looking at the repetitions of words (the first invocation on side B overflows to side A) suggests that the first side where the text starts is actually side B. The structure of the text is quite unique: After one or two long sentences (that I labelled introduction) a strongly repetitive part starts, perhaps 3 or 4 sentences long. The rich use of many special terms, conjunctions and pronouns in these repetitive parts inspired me to call them invocations (more on that later). This overall pattern is repeated two times - once on side B and once on side A. If we use a technique similar to what bioinformatics call "sequence alignment", the structure of the disc is even more easier to oversee:


This structural grouping reveals plenty of interesting features. First of all, there are the words marked with a 'wedge'. Markings of this sort were common in contemporary writing systems, for example, in Cuneiform Hittite, wedges ("Glossenkeil") were used for foreign words, like modern-day single or double quotes. In Egyptian, the ring-like cartouche marks were used to emphasize personal names, just like a capitalization. The only question: what was the objective of the scribe when putting these markups on the disc to stress certain words? It is obvious that the markings are associated to certain words and stems, but not specific positions in a sentence, so they are not phonetic or 'rhythmic' marks. Some words can apparently loose their marking if their grammatical case changes. This makes the phenomenon similar to certain modern languages, that capitalize proper names in writing, but not their adjectival or other derivatives. (English is a bad example, but the ortography of my native language, Hungarian, behaves exacly that way.)

The hint that these markings could signify proper names comes from the other corner of the Aegean. The Enkomi Cylinder, the longest Cypro-Minoan document discovered so far, shows a slightly similar feature. Though word-divisors (•) are commonly used on Linear A documents to divide words, they quite often divide phrases only, with individual words written continuously together. A similar dot can be encountered on the Enkomi Cylinder, but strictly and exclusively after certain phrases. The dots do not divide words in this case: rather, they are markups for some important terms, very likely names (for example, TA-LA-KA and SA-RA-NI, the latter being suspiciously similar to the Philistine term for 'prince', seren). This leaves us to ponder if the wedges on the disc are actually appropriately modified word-divisor dots, featured only after names. So the markups are likely equivalents of the Egyptian cartouche - used to stress proper names or any other important term, object or title that can be interpreted as a name.

Our second most important revelation is that the words on the disc show both prefix and suffix elements. As I expressed it in many posts before, the 'prefixes' are typically of a pronominal nature in Linear A, in fact attached particles and not grammatic prefixes. On the other hand, the disc also yields a number of suffixes, some of which we are already familiar with: like the -TE suffix (*-(a)the), used in the context of place-names in Linear A (as a type of locative or elative case). Its counterpair, the -QE suffix is fully novel. Judged by their consistent contrasting, I would suggest a locative-like sense of meaning to the -QE suffix as well. It is pure speculation, but the difference of meaning between -TE and -QE could have been based on something simple, like distance contrasting (-TE pointing afar and -QE pointing near). This would explain why the -QE suffix is never encountered on administrative tablets. But no matter how enticing, we clearly need to understand the text better before drawing such conclusions. Alternatively, this suffix could also mark an adjectival derivative or a different case.


The prefix-like pronominal elements are certainly the most interesting features of the text, not only because of their frequency, but also their uniqueness. No Linear A text has ever yielded particles I-QE-, MA- or SA-. Their use here can perfectly be explained by a change in context: While the tablets are strictly accounting texts, the Phaistos Disc is of a very different nature. This is hinted at by the similarity of the MA- and SA- particles (possessive pronouns?) to the well-attested Linear A verbal endings in -MI (1st person sing.) and -SI (3rd person sing.). The I-QE- prefix I already analysed in a previous post: it seems to be a simple conjunction (...and...), too redundant to be featured in the tablets' compressed context. Although likely derived from a deictic *i-, it carries the meaning 'and' instead. Such shifts in meaning are not uncommon in languages: for example, there is the Hittite conjunction nu-, literally meaning 'now', but typically used in the sense of 'and'. Another strange parallel between Minoan and Hittite we find here is the (stylistically perhaps bad) habit to frequently start new sentences with a conjunction ('and') in otherwise fully coherent texts.

The cited prefix elements can easily be read because of the fairly unique 'cat head' (MA) and 'linen' (SA) signs of Linear A. Others are trickier: for example, there is the bell-shaped sign Pha *07. It is undoubtedly a prefix-like element too, but not that easy to read. We can approach it from several direction. If we took a pure statistical (position and frequency) approach, we would get the best parallel in Linear AB *57 = JA. This is what Thorsten Timm already suggested, but there is a problem: the shape of Lin AB JA sign is so different from Pha *07, that a direct derivation is impossible. On the other hand, if we sought for similarly shaped signs in Linear A, we could get the sign Lin AB *37 = TI as the best match: and that is also a reasonably good match based on the terminal-heavy distribution of Pha *07. Linear A sign *41 = SI has a similar distribution too, but a somewhat dissimilar shape. But there is still a minor problem with this approach: Hieroglyphics have no bell-shaped signs at all. To overcome this problem, we can apply a third approach: searching for signs (in Hieroglyphics) that depict the same object as Pha *07. As the most probable object it could be is a female breast, the Hiero sign *34 (depicting a pair of breasts), clearly corresponding to Lin AB *59 = TA offers itself as a viable parallel. Now there is a different problem: the disc does have a sign (Pha *14) similar to Lin AB TA in shape, but obviously not depicting a pair of breasts!

To sum up, it seems that the disc contains first person references as well as numerous instances of proper names and phrases never seen on the tablets. I have a growing feeling that it is fitting to see the Phaistos Disc as an early example of poetry, a hymn very likely serving a religious purpose (as hinted at by the snake-shape, an attribute of a certain Minoan goddess). Religious hymns and prayers very often combine a rigidly repetitive structure with a high frequency of first and second person pronouns. As a simple example, let me cite the Lord's Prayer (Anglican/Catholic traditional version, the pronouns are highlighted in bold):

Our father who art in heaven,
hallowed be thy name,
thy kingdom come,
thy will be done
as on earth as it is in heaven.
Give us this day our daily bread.
And forgive us our trespasses,
as we forgive those who trespass against us.
And lead not us into temptation,
but deliver us from evil.
For thine is the kingdom, and the power
and the glory, for ever and ever. Amen.

On a quest to find Minoan pronouns

I could not steer myself away from the direction my previous post has started, even if I tried to. The Linear A records are becoming more-and-more fascinating with every little step we take towards decipherment. So we shall go straight to the next step towards understanding the Minoan language. It is time to turn our attention towards the Cretan pronouns.

This will be a difficult quest to start with. Now we face a bit different problem to that we did when searching for declensional endings. It seems that Cretan scribes were not only fond of leaving out endings, but they also tended to fuse short grammatical elements - such as pronouns - to words following them. So we are to see an awfully high number of enclitic-like forms, even when there was none in the original language. Now, to stop teasing, let us see all the putative pronouns we can find in Linear A!


Part I: Demonstrative pronouns

The easiest-to-find pronouns in Linear A are the A-TA- and TA-N- particles commonly occurring in the libation formulae, and other "longer" texts. Due to their high frequency and characteristic sentence-initial or phrase-initial position (compare TA-NU-NI-KI-NA [PL Zf1] with U-NA-KA-NA-SI [KO Za1]), one can relatively safely identify them as demonstrative pronouns. More precisely, the declensional variants of a single demonstrative pronoun with stem *(a)ta-. This word is not restricted to the Minoan language. Eteocretan inscriptions yield an evolved variant of this very A-TA- pronoun in the form "et": The phrase "et isalabre" from Dreros is fully comparable with the Minoan A-TA-I-*301-WA-E [PK Za11]. The TA-N- form on the other hand, is prominent in Eteocypriot inscriptions, as the following example shows: WI-TI-LE • RA-NU • TA-NA • MU-NO-TI • A-I-LO (text on a gravestone from Amathous). Looking further along the Mediterranean, one can discover that even in Etruscan, the demonstrative pronoun had the form of ta in nominative and tan in accusative cases. Maybe it is not mere coincidence, that most Indo-European languages also share a common demonstrative pronoun with a t-stem (sometimes reconstructed as *to). In Linear A, we only have two cases of this pronoun preserved:

Base form (nominative?)	Oblique form -n (accusative?)
at(a)-	tan-

Part II: Personal, possessive and relative pronouns

Evidence for the next group of pronouns comes from a number of intriguing phrases. For example, on a pithos from Zakros [ZA Zb3], we see the following formation: WINE 32 • DI-DI-KA-SE • A-SA-MU-NE • A-SE (....). The word-divisor dot after the third word is barely visible, so we might think of A-SE as an example of Suffixaufnahme - but only at the first sight. The E-A junction shows that the word A-SE is phonologically fully separate (otherwise it would have been A-SA-MU-NE-JA-SE, according to the junctional rules). Although A-SE is a frequently encountered term on the Haghia Triada tablets [HT81, HT93, HT132, A-SE-JA on HT115], a place-name (in nominative) does not fit the context here. Instead, it seems that this tiny little word A-SE repeats the ending seen on DI-DI-KA-SE. From our previous studies, we may know that this *-(a)se ending is a valid declensional case-ending (most likely a possessive type). Thus the stem of A-SE must have been a single *a-. Since pronouns tend to be very short in almost all languages of the world, the repetition of the suffix alongside with the extremely short stem makes this word suspicious of being a pronoun - most likely a personal or a relative pronoun.

And this is not the only example we have: on a libation table from Syme [SY Za2], we find another intriguing sentence: A-TA-I-*301-WA-JA • JA-SU-MA-TU • OLIVES • U-NA-KA-NA-SI • OIL • A-JA. In this example, the word A-JA exactly repeats the ending of the first, compound phrase. Anyone with just a little language talent would immediately sense a connection. The stem is - again - a bare *a, whereupon came the ending *-ja. So it seems, that we find another case of the same pronoun, in a similar situation.

Now it is time to remember our previous discussion about words that show an attached initial i- mysteriously wanishing in some cases. Other words have a- (or ja-, see later) initials, behaving in a fairly similar way. Good examples for *a- are rare but we can find pairs like TA-NA-TE [heading tablet ZA10], A-TA-NA-TE [ZA10, one line below] and SI-DA-TE [ARKH2, header] versus A-SI-DA-TO-I [ARKH2, one line below]. The addition and loss if initial *i- is much more common (see some of the examples here). We have already observed that the i- element might be interpreted as a short deictic particle. Now comes the surprise: this i- particle may also exist as a fully independent word in other cases. On the gold ring CR Zf1, the following text is written: A-MA-WA-SI • KA-NI-JA-MI • I-JA • ZA-KI-SE-NU-TI • A-TA-DE (After checking the original text, I corrected QA- on the fourth word to ZA-). The first two words should be familiar by now: these seem to be a combination of a personal name: "to *Armawa" (in some dative-like possessive case), and a first-person verb *kanijami : "I give" (KA-NI-JA- is a verbal stem and the -m/-n suffixes are used by a wide range of Eurasian languages to indicate first person singular of verbs). Because of these two words, a relative pronoun (referring to the ring) is pretty much expected to follow: and so we have I-JA. Because the fourth word looks like a second (compound) verb, a phrase like 'that-which' offers itself as a possible translation. (At the same time, the last short word might contain an *at(a)- demonstrative.) And this is not even the only instance to find the word *ija! The HT Zb158 inscription features the sequence (...) SU-KI-RI-TE-I-JA (on a pithos), where the unnatural joining E-I immediately wakes the suspicion that there are actually two words: SU-KI-RI-TE • I-JA (otherwise the joining would have been directly E-JA). The word SU-KI-RI-TA is also featured in a Phaistos inscription [PH Wa32]. Since this phrase is a terminal fragment of a longer text, a pronominal meaning for I-JA is - again - possible. If this is true, then we might have found another class of pronouns in Minoan: this time one with a simple *i- stem, behaving similarly to the *a- stem counterpart. A possible scheme for word-formation is summarised on the table below (remember that 'enclitic form' solely means that the word is never written separately: it might have been simply a writing convention):

Enclitic forms		Independent forms
basic form (nominative? article?)	oblique form with -n (possessive? accusative?)	with -ja ending (relative pronoun?)	oblique forms (declined)
a-	an-	aja	ase (possessive?)
i-	in-	ija	N/A (no example)

Having both an *a and *i stem for personal pronouns is reminiscent of the Etruscan language. Hittite and Luwian languages also display an enclitic third person pronoun with the stem *a-, though not in *i-. But unlike Etruscan, we have no evidence of animate-inanimate distinction between the two stems. The only difference seems to be the frequent usage of *i- in a deictic sense. This makes it similar to the Proto-Indo-European pronoun reconstructed as h₁e (Latin hic).

The *a- form of the first pronoun is only conjectural: none of the examples gleaned so far are decisive. If correct, these finds may reflect a 'definite article'-like usage of the base *a- and *i- forms. The *in- and *an- forms were recovered from compound sequences A-NA-TI-*301-WA-JA [IO Za8] and I-NA-TA-I-DO-DI-SI-KA [IO Za6]. The first sequence can most easily be explained as a result of a simplification (-TI- was written instead of -TA-I-), and an additional prefixed particle, *an- (corresponding to *in- in the second case). These *an- and *in- forms have two possible interpretations: First, they can be simple declensional cases (perhaps accusatives, as in the case of *ta-n), followed by the demonstrative *at(a)-. But the -n- consonant was also used as a pertinentive formative (*na), thus a possessive meaning cannot be ruled out. In the latter case, it is possible that the true pronouns we see here are in fact *anath and *inath = "in his" or "in its" (i.e. the so-far unattested pure locative case of a possessive pronoun). In Eteocypriot, the phrases A-NA • TA-SO and A-NO-TI • TA-SO-TI (again, from an Amathousian gravestone ) show that such formation is perfectly possible. The problem is, there are too few inscriptions discovered so far to decide this question at this time.


Part III: Miscellaneous pronoun-like elements

In this last chapter, I would like to discuss a single, yet enigmatic particle. This is the only case where we may probably speak of a true enclitic pronoun, since the particle *ja is not only attached initially (e.g. JA-TI-TU-KU [LA Zb1] vs. TI-TI-KU [HT35]) or terminally to words (this might be the very -JA suffix I mentioned in the previous post), but also inserted into the interior of words. To get what I mean, we shall analyse phrases KI-TA-NA-SI-JA-SE, A-NA-NU-SI-JA-SE and I-JA-TE.

Word KI-TA-NA-SI-JA-SE [PE Zb3] is one of the most baffling phrases written in Linear A. The -SI-JA-SE ending is repated in a different word: A-NA-NU-SI-JA-SE [HT Zb 159, on a pithos]. Since Hieroglyphic seals also feature a word KI-TA-NA, we can be practically certain that this is the stem, upon which all the formatives were built. It is also quite clear that the words end in an *-(a)se ending, that probably expresses an ablative case. The remaining -SI-JA- part is unlikely to be a separate word - but it is also too long to be a single suffix. While a suffix *-si (dative?) is possible to make out, it does not make sense at all, with a terminal -*(a)se. So we have to suppose there is a third part hiding in the chain (which is also probable on phonological grounds): *-ja-. If (and this is a big IF) this -ja- is an enclitic pronoun, then the entire phrase suddenly becomes meaningful: the second -(a)se ending in *kithana-si-ja-(a)se refers to a different subject, expressing something like: "to-Kithana-from-him". (the "him" is just a random pronoun I put in - we cannot be sure if it worked as a personal, demonstrative or relative pronoun).

Interpreting the *-ja endings as attached clitics does sound as a wild idea, or at least so I think. Yet (at least in some cases) it does offer a way to understand phrases written in Linear A. On a Phaistos jar [PH Zb4], in a fragmented inscription, we have the word I-JA-TE written out. The I-JA- initial is an uncommon one, resembling the phrase I-JA on the golden ring CR Zf1. Based on the context, it is certainly not ιατήρ = 'doctor' (this funny identification I found on Wikipedia). Its -A-TE ending rather shows a locative-type suffix. But if I-JA is a pronoun, with *i stem and *-ja suffix, why does it not use a regular locative form *ite or *jate or similar? One possible reason can be that this *ija is a case constructed from a suffix that isn't a true suffix. If the meaning of I-JA is a relativising phrase like 'that-which', could I-JA-TE mean 'that-from-which' ? (Since it is on a vessel, such a complicated interpretation may even be meaningful!) I know that there might be a simpler interpretation of *-ja endings: namely, a denominal suffix (if we disregard the examples with initial ja-/0 alternation and assume the stems *kithanas and *ananus). But who knows which version is correct?


To sum up, we have seen that - with our current level of knowledge - it is already possible to identify at least a few words or stems with pronominal function. Since they are really infrequent ones (we should not expect them to be common...), a good deal of work is left for the upcoming next generation of linguists who are to be delve deep into the secrets of the Minoan language.

A bold new theory on Minoan grammar - simple nouns and their compound suffixes

Back on the original track again, I decided to present a post on a theory I am currently working on. To undestand at least some details of the Minoan language is a really challanging task, but not challanging enough to deter me from making a post on a proposal regarding the grammar of the Minoan language. The task of decipherment is hard, due to a number of reasons. A large deal of this hardship is caused by the uncanny ability of Cretan scribes to abbreviate everything, including the total ommission of word-final consonants. In a language that frequently uses such tools (later we shall see that Minoan was probably such a language), the results are catastrophic: Even the affinity of the Minoan language is left uncertain. Any fluent speaker of the language could have relatively easily reconstructed the real meaning of the recorded phrases, but now we are facing tremendous difficulties, not knowing much about the speech of Bronze-age Cretan people. It is as if the correct English grammar would need to be reconstructed from a phrase like the following one (with respect to the elections, the quote is from Benjamin Franklin - the phonetics and ortography from Linear A):

E-NI • SO-SA-JE-TI • TA-TU-U-DU-KI-WA • A-RI-TI • RI-PE-TI • TU-KE-I • A-RI-TI • SE-KU-RI-TI • WI-DI-SE-WE • NE-I-TE • E-DU-RU-SO-PO

Although English is relatively poor in suffixes, we see that the verbs, articles and just about any other grammatically important part are essentially crippled in a script like Linear A. Although Hieroglyphics and Linear A were invented to write the Minoan language with, in many regards they were no better (or even worse) than the Egyptian Hieroglyphs (Linguists are still clueless about the vowels inside many-many ancient Egyptian words, even as Egyptian - Coptic - is still a 'living' language today).

Now, after all these warnings, let us turn our attention to the ancient Cretan scripts. Firt of all, I want to give you a sketch of a proposed noun declension system. I must remind all my readers that this is some sort of "preliminary work", something like the very first grids of Alice Kober - just scratching the surface and might not even be close to the truth. So, dear reader, treat everything you read on this page henceforth with extreme caution!


Part I : Declensional case endings

The simplest endings we can encounter on Minoan words consist of just one vowel. Since the Minoan suffixes are prone to intrude into the stem if they start with a vowel, these endings are expected to displace the orginal stem-vowel - in case the word ended with a vowel and not a consonant. To start with, I propose only two of these short or simple endings: one with -i , with a locative / dative / illative sense (at / to) and a counterpart -e with an ablative / elative sense (from / of). Since words could naturally terminate in -e or -i in nominative case as well, finding examples of these short endings is a rather hard task. This leaves this class of endings theoretic at best.

For the -i ending, we can encounter I-TI-TI-KU-NI on HT96. Since we can find numerous examples of TI-TI-KU on other tablets, it seems clear that this word was both prefixed and suffixed in the cited case. As for the initial I-particle, I have written about it enough in a previous post. But we also see a terminal -i here. As the Minoan Linear A syllabary cannot record terminal consonants, it is very probable that the original word ended with -n and the -i ending was added onto it (it was perhaps *thithkun - as another inscription [LA Zb1] shows a related word in the form JA-TI-TU-KU - note the way the consonantal cluster was resolved). For a word starting a header, an ending with a locative sense is the one most expected: in Linear B, most tablet-headers start with place-names, where the taxpayers or the goods stem from. For further examples, we can find place-names like I-DA-MI [SY Za1] (*i-tam(a)-i) and TE-KI [HT13], that occurs in the form TE-KE on tablet HT85. The two latter names are almost certainly referring to the same place, because other place-names coincide on the two tablets.

The existence of -e endings is little more than merely conjectural: in the libation formula, we have seen phrases like (J)-A-SA-SA-RA-ME, that aren't exactly in the nominative case. In our previous posts, I left that phenomenon unexplained: but now we may identify it as an -e ending, with a sense of "of *jasasaram(a)" Other tablets supply us with further words suspicious of belonging to this class: for example, there is the variation of A-TA-I-*301-WA-JA/-E. But again, the high number of words terminating in -e that are not of this case (probably being nominatives) hampers our work.

Even though the "simple/short cases" above were little more than merely conjectural, they are actually of great help explaining the system of the long/extended cases. These long or extended cases seem to be formed with the junction of a consonantal suffix (occasionally with a helper vowel) and the previously-seen vowel-based suffix: in other words, they are either of a -CV or a -VCV type. As there are three possible consonantal bases, the proposed system of extended suffixes would look like the following:

Consonantal part	+i (to/at?)	+e (from?)
-(a)s (possessive?)	-si (to/at the possession of?)	-(a)se (from the possession of?)
-(a)th (locative?)	-thi (to/at the location of?)	-(a)the (from the location of?)
-(a)l (pertinentive?)	? (not attested)	-(a)le (from the family of?)

The list of examples for these extended suffixes is much more longer than that of the short ones. It is not always simple to discern them though. For example, the -si and -ti endings can also occur on verbs. Fortunately, in many cases, the context can be helpful to determine the most probable meaning.

A good example for the possessive series is the word-pair PI-TA-KA-SE [HT21] (where it stands as a donor of goods) and PI-TE-KE-SI [HT87] (heading a list of people, and the term may specify the location or receiver of workforce). The -SI anding can also be caught in the term O-KA-MI-ZA-SI I-NA on tablet PK1. The highly unusual I-I junction shows that this term is actually two separate words: the first one carrying the ominous possessive suffix -si. The same ending is seen in words like E-NA-SI [KH7], U-TA-I-SI [KH16] and even ?-RA-MA-SI [ZA11], written on a tablet mentioning outgoing goods. From further examples, such as the golden ring CR Zf1, which features the sequence A-MA-WA-SI • KA-NI-JA-MI or the tablet HT28 with the header A-SI-JA-KA • U-MI-NA-SI on side B, it seems more-or less clear that this -si suffix does not simply express possession (like a genitive), but rather stands in a sense of dative. On the other hand, the -(a)se ending occurs frequently on the pithos-type jars found in the cellars of temples, and also on tablets listing taxpayers, like A-NA-NU-SI-JA-SE [HT Zb159], KI-TA-NA-SI-JA-SE [PE Zb3], O-TA-NI-ZA-SE [ZA5], DU-RE-ZA-SE [ZA10], TU-MI-TI-ZA-SE [ZA14], MI-ZA-SE [ZA15], SO-KE-MA-SE [ZA15] or RU-MA-TA-SE [ZA20]. We can also find word-pairs like RU-MA-TA [HT29] and RU-MA-TA-SE [ZA20]. The best example of this suffix is DI-DI-KA-SE on ZA Zb3. In this case, the context (especially the presence of a passive verb like A-TA-I-*301-DE-KA) strongly suggests an ablative sense of usage ("from the possession of...").

For the use of locative endings, like the -thi suffix, a great example is the pair RI-RU-MA [HT118] / RI-RU-MA-TI [PH31]. Another one consists of DA-KU-SE-NE [HT103] / DA-KU-SE-NE-TI [HT104]. Its counterpair, the -(a)the suffix is found in words like I-DA-MA-TE [AR Zf1] (despite my earlier interpretation, it could also mean "from the sanctuary"), or I-JA-TE [PH Zb4] (on a pithos sherd, possibly meaning "from that"). One can also add the example of the pair KU-PA3-NA-TU [HT47, HT119 tablets] / KU-PA3-NA-TU-NA-TE [AP Za2, on a stone vessel], where the original nominative of the noun - a place-name - was perhaps *Kurphan, ending in -n. Unlike -thi (that are more like regular locatives), the -(a)the endings are supposed to express an ablative or elative-like meaning ("from the place of..."). Such an interpretation of the -A-TE suffixes is not novel: it is in full accordance with the earlier theory of Miguel Valério on -TE suffixes (i.e. that of JA-DI-KI-TE-TE DU-PU2-RE).

The -(a)le class, as its pertinentive meaning suggests, occurs almost exclusively in personal names. The list is very long, and includes examples like A-RA-NA-RE [HT1], I-NU-MA-RE [ZA4], JA-MI-DA-RE [HT122], MI-RU-TA-RA-RE [HT117], TE-JA-RE [HT117] and WA-TU-MA-RE [HT128]. We find the same form in DU-RA-RE [KN Zc7] and O-SU-QA-RE [TL Za1] - a recurring theme on the libation tables. Another very nice example is the header of a tablet from Petras [PE1]: U-KA-RE • A-SE-SI-NA • KU-PA-RI. The first word of the three seems to carry this very -(a)le ending. The table lists the number of men alongside their food supplies, and the stem U-KA- is suspiciously similar to the phrase used in Linear B for "troops": O-KA. Thus the interpretation of the -(a)le ending as a pertinentive case-marker is fully meaningful. While the second word on the same tablet seems to be an adjective of -na ending (more on this later), the last one presents another example of an -i type ending.

I must mention though, that for the above system to work, one must assume a fully hypothetic class of medium-length or consonantal suffixes: that is, the pure consonantal endings [-(a)s, -(a)th, -(a)l]. The existence of this class is impossible to prove on the basis of Linear A texts: the Cretan writing conventions dictated that terminal consonants were always ommitted (unlike the later Cypriot Linear C). One can only guess at their existence where an unexpected vowel-change occurs, due to the addition of the auxiliary vowel onto the stem. A possible (though uncertain) example could be the strange formation observed in the I-PI-NA-MA / I-PI-NA-MI-NA pair: if the original word ended in -i, the -a ending on the first version could be the result of an intruding suffix, such as *-as.

It could be perhaps added, that no one was ever able to identify any clear plural marker for nouns in the entire Linear A corpus: therefore there is a distinct possibility that the plural marker was a purely consonantal-type suffix, always being ommitted from the records. As a rare example, in the compound U-NA-(RU)-KA-NA-SI/-TI, we find a -RU- extension on the noun corresponding to a putative plural marker. Yet caution is advised: there is no other known example of a -RU ending as plural, so it is probable that this ending is a derived one (not a simple nominative case) - in which case the plural marker could have been the naked -r- part (*una-r-u+?). Alternatively (though less probably, since we also have U-NA-A [KN Zb40]) the stem of this word could have contained this very -r-; and in this case the missing plural marker is only traceable because of the auxiliary vowel -u- it used to fit onto the stem (*unar-u+?).


Part II: Suffixes of unknown affinity

I must admit, there exist some suffixes that cannot be assigned a clear meaning without any contraversion. Apart from hapaxes, I could so far identify two of such endings: the A-A and the -JA suffixes.

The first of these suffixes is the very rare and mysterious -A-A case. It is very easy to discern (as an A-A junction is never normally observed outside this case). Problem is, there are only two known occurrances of this case-ending: U-NA-A [KN Zb40] and I-DA-A [KO Za1]. Though clearly a marker of some unidentified grammatical case, the sense of its use cannot be deciphered from just these two cases. It is also possible, that the real suffix was longer, and ended in a consonant we may never know.

The second suffix of uncertain affinity is the -JA ending observed on some words. It can be - at least theoretically - a simple nominative suffix. But problems abound with this interpretation once we turn to the Minoan pronouns (more on this matter in a later post - it deserves its own). All we know is the -ja ending sometimes alternated with -e (the latter expressing some ablative-like case). The use of this ending even persisted into Eteocretan (compare the words isalabre [*isal-awr-e] and isaluria [*isal-awr-ja]). The most important Linear A example is the alternation of A-TA-I-*301-WA-JA [PK Za12] and A-TA-I-*301-WA-E [PK Za11]. Other possible Liner A examples of this -JA suffix include the pair KU-PA [HT110] / KU-PA-JA [HT116] and PA-SE [HT18, HT27] / PA-SE-JA [HT93, HT Wc3001]. As there is no hint that the language behind Linear A was an ergative one, I could not make a good sense of this ending so far. It might have been some "clitic" as well, implying stress on a particular noun, irrevocably fused to the stem on some words, especially on those that were later borrowed into Greek (e.g. in the case of theonyms like Εἰλείθυια).


Part III: Derivative suffixes

The last class of nominal suffixes observed in Linear A or on Minoan words inherited by Greek is that of the adjectival formatives. We have evidence of three different classes. The first one: -na is well attested in Linear A (e.g. JA-SA-SA-RA-MA-NA = *(j)asasaram-na) as well as in place-names like Mycenae (*Muka-na). Of course, it could have been combined with the above-seen *-e suffix to yield the very common -NE ending of personal names on Linear A tablets. The next one is the famous -issos/-issa class of Pre-Greek names - it may correspond to the part -I-ZA or -E-ZA (-*(i)za with affricate?) found terminally (or sub-terminally) in a number of Linear A words, especially names. Take the example of DU-RE-ZA-SE [from ZA10]: *tule-(i)za-(a)se. Due to their nature, the formatives typically precede declensional case-endings (if there is any). Last, but not least, there is the case of Pre-Greek names in -inthos/-intha. The latter one may derive from a Minoan '-I-TA' -like suffix, with prenasalisation, of a form *-(i)ntha. The meaning of all three classes are subtly different. I promise to write a full post on them later, but beforehand, I mention that they are expected to express the following flavours:

Pertenitive adjective	Genitival adjective	Collective adjective
[stem]-na	[stem]-(i)za	[stem]-(i)ntha
(with [stem] ?)	(of a single [stem] ?)	(of multiple [stem] ?)

Now, to put our theories to the test, I shall post three different Linear A table-headers we will attempt to interpret with our newly-gained knowledge:

Number #1 (tablet HT117):

MA-KA-RI-TE	KI-RO	U-MI-NA-SI	[names with '1']
from place *Makri or *Makrintha (place-name?)	missing (past pariciple of verb *kir- = "to be missing")	for *umina (sacrifice?) (cf. Etruscan *umi- = "to dedicate")	[names of people required to pay "sacred" service]

Number #2 (tablet HT87):

QI-TU-NE	MA-KA-RI-TE	PI-TA-KE-SI	[number '1']
(noun): servant? qualifier of men?	from *Makri(ntha)(place-name?)	to *Pinthake (personal name?)	[amount of workforce allocated]

Number #3 (tablet PE1):

U-KA-RE	A-SE-SI-NA	KU-PA-RI	[men and grain]
of the troop? (cf. Linear B O-KA = 'troop')	(adjective): 'allocated' vel sim?	to *kupa(r) (*-i ending? *-r- formative? plural?)	[quantity of men (soldiers?) and their food supplies]

As the above examples show, the theory is a viable one, yielding a reasonable meaning for most of the phrases seen on Linear A documents. Still, we are far from having a proven and accepted theory on our hand, but at least we tried to construct a consistent one. I am sure this one will need a lot of improvements, even if it is basically correct. It is also imperative to look at the grammar of other "Aegean" languages, such as Eteocypriot, Lemnian and others. Apart from undeniable similarities, there are also notable differences: for example, the "duality" of extended (agglutinated) cases is not observed in Etruscan, yet it clearly seems to be a feature of the Minoan language. Whether this reflects the ancestral structure of Aegean languages, or just a side-branch development specific to Minoan, or rather that the theory needs amendment, I have no idea. But - so I hope - future will tell.