Peter van Soesbergen thinks he's deciphered Linear A.
The script shares the majority of its signage between A and B. We can read B as ancestral to Arcado-Cypriot Greek. Since A and B coëxisted in Crete, several placenames coïncide, notably pa-i-to > Phaistos. The libation vessels also preserve transcriptions, like qa-qa-ru > qa-qa-ro. Thus the temptation not just to read Linear A aloud, like we do for Etruscan; but to translate it.
Cyrus Gordon thought A was Semitic; Frederik Wuidhuizen thought it Anatolian, specifically Luwian. These languages were spoken in ship-reach to Crete, both easterly - which is where the Middle Bronze Age palaces oriented before 1600 BC. Gordon could point to Semitic-like terms in the Linear A ledgers: compare ku-ro to (say) Arabic kull. Wuidhuizen could compare southwest-Anatolian nonGreek place names like Tarhuntassa - with its Luwian possessive suffix - with Cretan placenames from the centre-east, like Knossos.
Somewhere around here are claims that Linear A is a wayward Greek dialect, like Pictish is some kind of Celtic. Remember pa-i-to? It's not "pa-i-tu". Linear A didn't much use *o-grams but it had them.
Gordon's problem was that technical-jargon may wander across scribal communities, cf. English "total". Wuidhuizen's problem is that the Luwian suffix is not shared in Hittite, which has a genitive like protoIndoEuropean and indeed Greek; so Luwian probably borrowed its suffix. Likewise borrowed could well be the libation-formulae, as Catholics once prayed in Latin and indeed the Hittites were praying in a dozen languages. A problem for err'body is that Linear A/B synoptic placenames cluster west. Oh right, and before the 1700 BC reforms and great feasts, Cretan foreign trade was going south, to mighty Egypt (and the Qeheq?). And what shackles one script to one language? Anybody could write cuneiform, like I'm writing English in a miniscule Latin alphabet.
The ~1600 BC Thera tuff / tephra blew to the southeast, as it happens; so perhaps we should be unsurprised that surviving Cretans had to shift to the poorer west. There, I think, the Greeks were waiting. Surviving Linear A, mostly clay, looks inscribed between-days, when Knossos was weak. Phaistos was then stronger: most Linear A is from nearby Hagia Triada. So we get lots of Pa-i-to in Linear A, but Knossos is difficult to ascertain (Ku-ni-sa?). Of course in Linear B, Ko-no-so is king of Crete again, but Greek now. (The Hagia-Triada palace is Linear B Da-wo, Linear A Ka-pa.)
A key assumption to van Soesbergen, is that the Cretan brand of Arcado-Cypriot didn't distinguish aspirants. So ka-ko (copper, bronze) wasn't just a naming-convention for our classical Ionic khalkos; 'twas truly *kalkos: on its way to khalkos, or the latter existed in dialects only outside Crete (many say that Doric already existed on the mainland). Sounds begadkepat to my ear, what will happen to Hebrew and Aramaic.
With that assumption, 1500s BC Cretan Greek lacked kh-. What this "Kalkolithic" Greek did have, from the IndoEuropean days, was the "labio velar": qw-. Van Soesbergen questions that Linear A had the labio-velar; Semitic Q is instead emphatic-K oft-writ "Ḳ", and Greek itself would lose its Q later. He posits that something else in Linear A got repurposed, in Cretan Greek "B", to transcribe that qw-. Linear A "qa-qa-ru", then, was not "quaquaros" except for Greeks.
Van Soesbergen elsewhere points to the common ending to personal-names, -te-ja. This -teya is shared with Hurrian cuneiform, where it swaps out -teshub as a "hypocorism" that is, dimunitive (*Alexandros > "Sasha"). He restores Gordon's reading -ja here where some have read -pi.
Elsewhere-elsewhere, van Soesbergen finds that Linear A is agglutinative and ergative. There are elements of these in Sumerian; but why would Crete use Sumerian on clay without the cuneiform? Nah: this guy sees here instead, aspects of Hurrian. He would reconstruct the B qw- in Linear A: ḫ-. Which is rife in Semitic and Hurrian; and indeed maintained in Anatolian (but not Greek) from IndoHittite h2.
Van Soesbergen has the benefit of doing a lengthy transcription of every Linear A text he can find; and he doesn't shake the tree too hard for previous readings of the script, besides that admittedly-drastic q/ḫ switchup. I would take it seriously, which I never did for Gordon. Although honestly, whatever we think of Wuidhuizen's bombast, scholars also rate Finkelberg highly. The Internet notes that van Soesbergen simply hasn't addressed Finkelberg's argument, which bodes ill for him.
