Proto-Elamite and Linear Elamite

The 2022 decipherment has been tested so far by reception rather than by formal replication. Matthew Stolper, Piotr Steinkeller and Manfred Krebernik judged it coherent and plausible, Krebernik with the proviso that it be right in every detail; Massimo Vidale called it a major discovery⁠[8]. The Bern group, whose OCLEI corpus and earlier syllabary work were independent of Desset, lists a forthcoming paper offering "external and semi-external sound value confirmations" of the 2022 values, which is the nearest thing to an independent check⁠[9]. The obvious weaknesses are also on record: the texts are short and formulaic, so the readings rest heavily on names; the silver vessels have no excavation record and only a reported origin; and the authors themselves say that Elamite is imperfectly understood, so translation stays difficult even where the signs are read⁠[6]⁠[8]. Metallurgical analysis is reported to have found the vessels ancient rather than modern⁠[8]. Dahl's objection has hardened: in 2025 he called the 2022 paper "so full of errors" that it should not have been published, and some scholars object to building a decipherment on objects from irregular excavations⁠[25].

The continuity claim has been tested against the corpora. Dahl argued in 2009 that Linear Elamite's signs are unstandardised and sometimes pseudo-glyphs, and that its Proto-Elamite look-alikes are borrowed, not inherited; Dahl and Kelley repeated the 800-year gap objection in 2022⁠[11]⁠[8]. Mäder's 2020 comparison found 50.5 per cent of sign forms identical between the two scripts and a further 16.8 per cent similar, and picked out seven inscriptions that share features of both; the Bern group also reports a new inscription from Shahr-e Sokhte dated about 2600 BC that would narrow the gap⁠[26]⁠[9]. Neither the comparison method nor the Shahr-e Sokhte text has been tested by others in print.

Proto-Elamite has been the subject of a run of machine-learning papers from 2019: sign clustering and topic models reproduced known groupings and proposed new ones; later work on compositional graphemes, numeral disambiguation and image-aware language models aimed at revising the sign list rather than at reading⁠[22]⁠[23]⁠[1].

On Linear Elamite, three positions. The first is that Desset and colleagues have read it: a purely phonographic script writing Elamite, with 72 sign values that hold across the corpus and give sensible dedications and royal names⁠[7]. The second is that the core is right but the system is mixed: Mäder's statistical analyses suggest only about 70 per cent of signs are phonographic and the rest logograms, which would change some readings⁠[5]. The third is that the values are not yet established, because the texts are short and formulaic and the decisive new material is unprovenanced, so the readings should be held until a longer text or an excavated bilingual confirms them⁠[6]⁠[8].

On Proto-Elamite, the live question is whether it encodes language. One view is that it writes an early form of Elamite with a mix of logograms and syllabic signs, which Desset's continuity argument requires and which the recurrence of sign sequences in personal-name positions would support⁠[7]⁠[8]. Another is that it is an accounting notation that recorded quantities, commodities and perhaps names without encoding speech, so that its "decipherment" would be the recovery of the bookkeeping system rather than of a language; the absence of any positive evidence for Elamite and the purely ideographic earliest tablets support this⁠[1]⁠[3]. A third holds the language question open and treats the sign list as the only tractable target⁠[23].

For Linear Elamite, an excavated text long enough to contain grammar, or a new bilingual from a documented context, would test the 2022 values outside the material they were built on; so would independent readings of the same vessels by scholars working from the OCLEI corpus with the values fixed in advance⁠[24]⁠[9]. The full edition with translations has begun: volume I (the Susa texts) appeared in 2026, and volume II, with the silver vessels on which most readings rest, is the one to check⁠[10]⁠[20]. For Proto-Elamite, a stable sign list published with the CDLI corpus, whose bulk export already makes every transliteration downloadable⁠[27], followed by positional statistics of the non-numerical signs against the expectations for a syllabary, a logography and a non-linguistic notation, would show which it is; the image-aware models are a step towards that list⁠[23]⁠[22]. The continuity question turns on dated intermediate inscriptions, such as the reported Shahr-e Sokhte text⁠[9].