Eteocretan and Eteocypriot

The Eteocretan corpus is six stones and a few fragments. Three come from the Altar Hill sanctuary at Praisos and are in the Heraklion museum: Praisos 1 (inventory 99), five lines in archaic Cretan letters written boustrophedon, found by Halbherr in 1884; Praisos 2 (inventory 100), thirteen lines in the Ionic alphabet of about the 4th century BC, found by Bosanquet in 1901; and Praisos 3 (inventory 101), sixteen lines in Ionic with digamma, of about the 3rd century BC, found in 1904⁠[22]⁠[23]⁠[11]. Two come from the cistern beside the temple of Apollo Delphinios at Dreros, found in 1936 and published in 1946: a grey schist block 75 cm wide with two lines of Eteocretan above three of Greek, and a second stone of which only the Eteocretan word tuprmeriei a and its Greek survive⁠[10]. The longer Dreros stone was kept at Neapoli and lost during the Second World War⁠[24]. The whole corpus is about 420 letters⁠[2]. The standard edition is Duhoux's L'Étéocrétois of 1982, and Ray Brown's web pages give every text with dimensions and bibliography⁠[1]⁠[11].

The Eteocypriot corpus is larger but shorter per text. Steele catalogues 26 inscriptions, EC 1 to EC 26, nearly all from Amathus and most of them short dedications or funerary texts of the 8th to 4th century BC⁠[3]⁠[4]. The Amathus bilingual, a black marble slab broken in two, was bought in a Limassol antique shop and published by Sittig in 1914; it was lost soon after, possibly in shipment to America⁠[12]. It is ICS 196 in Masson's corpus and no. 7 in the new Inscriptiones Graecae XV 1.1 of 2020⁠[25]⁠[26]⁠[13]. Among several hundred Cypriot syllabic inscriptions that cannot be read as Greek, only those with clear non-Greek morphology are counted as Eteocypriot⁠[3].

The scripts are not in doubt. The Praisos and Dreros texts use the Cretan alphabet of their Greek neighbours, so every letter has a known value, and Duhoux argues from the letter forms that the Eteocretans took up the alphabet at the same time as the Greeks of Crete⁠[27]. The Cypriot syllabary has 55 to 56 signs for open syllables and bare vowels, runs right to left, and writes a final consonant with a dead vowel; it was deciphered for Greek between 1871 and 1876 and is read with only a few rare signs uncertain⁠[28]⁠[6]⁠[3].

What the texts say is known only at the edges. In the Dreros bilingual the Greek has ewade, "it was decided", and turon, "cheese"; van Effenterre matched inai in the Eteocretan to ewade, and the sequence phraisoi inai in Praisos 2 would then be "the Praisians decided"⁠[10]⁠[11]. The city name appears as phraiso-, with an aspirate the Greeks did not have⁠[1]. Word dividers on Dreros 1 and Praisos 1 show words such as et, komn, men, lmo and barze; clusters like komn and lmo suggest syllabic n and l, and the digamma between t and s in Praisos 3 points to a sound Greek lacked⁠[11]⁠[1]. Recurring elements -mit- and irer suggest morphology but no paradigm can be built from 420 letters⁠[2].

For Eteocypriot the bilingual gives the names Ariston and Aristonax in the forms a-ri-si-to-no-se and a-ra-to-wa-na-ka-so-ko-o-se, which shows that the signs keep their Greek values and that -o-ko-o-se marks the patronymic; Steele finds this patronymic, usually on Greek names, the one secure diagnostic of the language⁠[3]⁠[4]. The ending -o-ti recurs and has been compared with a Cypro-Minoan sequence⁠[29]⁠[20]. The Eteocypriot line is longer than the Greek, so some of it has no Greek counterpart and the equation a-na ma-to-ri with "the city" is a guess⁠[3].

Every reading has been tested against the same two things, the bilinguals and the internal recurrences, and none has passed both. Davis's 1961 Anatolian reading of Praisos and Psychro, built on Hittite look-alikes, was rejected in review for assigning meanings freely to isolated words⁠[14]. Gordon's Semitic reading from 1962 took lmo as "for his mother" and found Semitic roots in Praisos; Duhoux and Brown object that the matches are single words without the inflection Semitic requires, and the 2016 revival by Magnelli and Petrantoni rests on one word at the end of the Dreros text⁠[16]⁠[1]⁠[11]⁠[30]. Brown's and Facchetti's Tyrrhenian grouping depends on a handful of endings shared with Etruscan and Lemnian, which Rix and De Simone regard as insufficient⁠[17]⁠[18]⁠[24]. Duhoux's own analysis of 1982, which worked only from the texts, found a language that is neither Greek, Indo-European nor Semitic, with hints of agglutination, and stopped there⁠[1].

The Psychro stone, five alphabetic words with three Linear A-like signs, was accepted as Eteocretan by Brown and Duhoux but argued to be a modern forgery by Kritzas in 2006; the Sumerian-based decipherment that Kenanidis and Papakitsos published in 2017 therefore rests on a text most epigraphers no longer admit⁠[31]⁠[32]⁠[33]. For Eteocypriot, Jones's 1950 segmentation and Petit's 1999 Hurrian proposal from names and toponyms have not been confirmed by any reading of a text, and Steele's 2013 review of all 26 inscriptions concludes that no secure link to a known language can be shown⁠[34]⁠[19]⁠[3]. Duhoux's 2010 comparison of Eteocypriot endings with Cypro-Minoan 1 sequences, and Janko's 2020 reading of the Enkomi cylinder as an Eteocypriot account, are consistent with continuity but cannot be checked while Cypro-Minoan values are themselves unproven⁠[20]⁠[21]⁠[35]. No computational or statistical study of either corpus has been published; at 420 letters the Eteocretan corpus is below the size at which frequency methods discriminate between languages⁠[2].

For Eteocretan five positions are live. The first is that it is an Anatolian language, as Davis argued from Hittite and Finkelberg from Lycian, fitting the view that pre-Greek Crete was settled from Anatolia⁠[14]⁠[15]. The second is Gordon's Northwest Semitic reading, revived in 2016 for one word of the Dreros text⁠[16]⁠[30]. The third is the Tyrrhenian or "Aegeo-Asianic" grouping with Etruscan and Lemnian proposed by Brown and developed by Facchetti and Negri⁠[17]⁠[18]. The fourth, held by Duhoux and most epigraphers, is that Eteocretan descends from the Minoan language of Linear A and belongs to no family with other written records, so that only internal analysis and new texts can advance it; Duhoux warns that the thousand-year gap means the two must still be studied separately⁠[1]⁠[24]. The fifth is that the corpus is too small for any of these to be decided.

For Eteocypriot the positions are fewer. Petit's Hurrian hypothesis rests on Amathusian names and the island's Bronze Age contacts rather than on readings⁠[19]. Duhoux, Valério and Janko argue that Eteocypriot is the Iron Age survival of the language, or one of the languages, written in Cypro-Minoan, which would make the Amathus texts the only alphabet-adjacent key to that script⁠[20]⁠[35]⁠[21]. Egetmeyer adds that the Golgoi texts may be a different non-Greek language, "Golgian", so that there may be two unknowns rather than one⁠[4]⁠[36]. Steele's position, the default, is that Eteocypriot is an unknown language whose morphology can be described but whose relatives cannot be named⁠[3]⁠[37]. No one argues that either language is Greek.

The Eteocretan question needs text. Praisos was a large city occupied to 145 BC and only its Altar Hill has been dug for inscriptions; a longer bilingual decree, or even a few more lines with word dividers, would let the recurring elements be tested as case or verb endings⁠[23]⁠[2]. Two pithos handles from Azoria stamped with a non-Greek sequence show that stray finds are still possible⁠[2]. Short of new finds, a proposal must read all six stones with one grammar and account for the Greek of the Dreros bilingual, as van Effenterre's cheese regulation tries to; a reading that fits one word and ignores the rest fails the test that Duhoux set in 1982⁠[1]⁠[10].

The Eteocypriot question has two routes. One is more bilinguals: Steele counts four Greek and Eteocypriot texts already, and Amathus is still being excavated, so a bilingual long enough to align clauses rather than names is possible⁠[4]⁠[3]. The other is Cypro-Minoan: if values for its signs can be fixed on Cypriot texts and tested blind on the Ugarit tablets, the Amathus words can be sought in the Bronze Age corpus, and the -o-ti and -o-ko-o- endings become a test of continuity rather than a suggestion⁠[20]⁠[35]. In both cases the Linear B parallel sets the standard: Ventris's values were accepted because they produced Greek words that fitted the tablets' logograms and place names, a check from outside the system⁠[8]. For Eteocretan and Eteocypriot no such outside check exists yet, which is why the sounds can be read and the words cannot.