Here is the peculiar fact at the centre of Linear A: we can pronounce most of its signs without understanding a single sentence. You could hand a tablet to a specialist and they would sound it out — and then have no idea what they had just said.
We can do this because the signs were later borrowed to write early Greek, in the related script called Linear B, which was deciphered. Reading the sound-values back into Linear A is a careful reconstruction — never fully certain, but disciplined. What the reconstruction does not give us is meaning. Sound and sense are two different problems, and only the first one is solved.
So after more than 120 years of study, no one has identified the language behind Linear A with any confidence. The few words we trust come from watching how they behave on the tablets, not from cracking the language: ku-ro “total” and ki-ro “deficit” sit where a total and a shortfall would sit on any ledger. Beyond a handful like these, the language is open.
Two answers can be set aside with confidence. Linear A is not Greek. Greek arrived in the Aegean later; when Mycenaean Greeks adapted this script for their own language, the fit was awkward, the way a borrowed alphabet usually is — a sign that the system was built for something else first. And it is not Egyptian, despite Crete’s trade with Egypt; the script and the language behind it show no such relationship.
That still leaves an enormous space. Ruling out two candidates is not the same as naming the language — it only narrows where to look, and the remaining field is wide open.
There are two very different ways to attack an unknown language, and they do not have equal standing. The respectable one is the combinatorial, or contextual, method: you study the script’s internal patterns — which signs cluster together, where a word sits in a list, what numbers follow it, whether it heads a tally or closes one. This is exactly how ku-ro was pinned to “total”: not by what it sounds like, but by where it always appears.
The tempting one is the etymological method: take a Linear A word, notice it sounds a bit like a word in some known language, and declare that the meaning. It is seductive because it produces instant, vivid “translations.” It is also how most amateur decipherments go wrong.
Davis (2026) explicitly warns against the etymological method as circular reasoning — you assume the language family to find the word, then point to the word as proof of the family. With enough languages to fish in, you can “match” almost anything to almost anything. A resemblance is a hypothesis to be tested, never a result. On this site, any meaning that rests on resemblance alone is treated as unproven.
Candidate relationships have been proposed over the years. Two come up most often. One links Linear A to the Anatolian languages — the family that includes Luwian, spoken across the water in Bronze Age Asia Minor. Another links it to a Semitic language, on the strength of a few resemblances in counting or commodity terms.
Both are worth stating plainly: these are proposals, not findings. Linear A has not been shown to be Luwian, nor Semitic, nor a member of any known family. Each proposal leans heavily on the resemblance-matching that the section above flags as unreliable, and none has produced a reading that holds up across the corpus. Treat “Linear A is X” — for any X — as a claim that has not been demonstrated.
What scholars are willing to say about the language’s shape is far more cautious, and offered as hypothesis. Some, such as Duhoux, have suggested the language may build words by adding pieces to a stem — a tendency sometimes described as prefixing or agglutinative. That is a structural impression drawn from internal patterns, not a proven grammar, and certainly not an identification.
Part of why this is so hard: the Minoans were on Crete long before Greek-speakers arrived in the Aegean. Their language belonged to a world that the later Greek-speaking one largely replaced — and it has no identified living descendants. There is no modern tongue we can hold up beside it, no related language to lean on, the way Greek lets us read Linear B.
So the honest bottom line is a strange one. The script is readable aloud; the language is open. We can date these tablets to within a few centuries, sound out their words, and recognise a “total” at the foot of a list — and still not know what people the writers belonged to, or what most of their sentences meant. That is not a gap waiting to be filled by the next clever guess. It is the current, careful state of the evidence.