The Ancient Greek interrogative ἆρα is strikingly similar to modern Persian āyā.

Both words exclusively signal yes/no questions, and almost always begin the sentence. There is an accent on the first syllable in āyā, similar to the circumflex in ἆρα.

āyā is at least as old as modern Persian itself: it occurs in the epic Shahnameh (c. 1000 AD), with the same form and function. Quite possibly it comes from Middle/Old Persian.

api in Sanskrit seems to play the same role.

Are these words cognates? Are there other Indo-European languages with a similar interrogative particle? (French uses est-ce que, but that's an entire phrase.)

In other words, do ἆρα, āyā, and api share a PIE ancestor? Did it survive in any branch other than the Hellenic and Indo-Iranian? I would appreciate any clues.

  • cp E uhu?, ah?, oh yes?, G ah ja, ach!, ach ja!?!; G *naja, Fr ouais; OE a, E ever, however, anyhow etc; also right vs just
    – vectory
    Commented Dec 11, 2019 at 13:42

3 Answers 3


ἆρα is considered to be cognate with the interrogative particle in Baltic languages (Latvian ar, Lithuanian aȓ).

Persian āyā does not have a known ancestor in Old or Middle Persian. In early New Persian it does occur, alongside ay اى with the same function. From the viewpoint of regular sound correspondence, any connection with ἆρα is quite out of the question. Persian -y- cannot derive from IE -*r-.

  • 2
    IE full grade *h2er-, zero-grade *h2r-, has no known cognates in Indo-Iranian. In Greek the full-grade stem gives ἄρα, shortened to ʼρα. ἆρα is from ἤ ʼρα.
    – fdb
    Commented Jan 5, 2017 at 17:33
  • 1
    The connection with Latvian/Lithuanian ar is not necessary and the Baltic particle can be easily explained solely on Baltic grounds. The most economical solution is to consider Greek ἆρα as as Greek problem as well. To connect these words, with such an enormous attestation gap with no additional evidence from other subgroups is quity risky. Not to mention extremely unstable status of a particle as a word-category.
    – czypsu
    Commented Jan 14, 2017 at 14:48
  • 2
    @czypsu Beekes etymological dictionary mentions the relation to the Baltic particles. Specifically he says: *Related to Lith. ir̃, Latv. ìr 'and, also; even' and (with full grade) the question particle Lith. ar̃, Latv. ar. * As I noted below Phrygian has it as well, so it is not a Greek problem.
    – Midas
    Commented Jan 14, 2017 at 18:15
  • 1
    I didn't say that the theory is non existent, I just recently spoke with a historical linguist with great experience in Baltic who presented a really convincing Baltic explanation of ara. And Beekes, being a very good scholar, is not always to be believed unconditionally (e.g. his Pre-Greek reconstruction). Moreover, in my view, the whole idea of reconstructing a particle for PIE can be ill-advised seeing how unstable, functionally and formally diversified it can be. Supporting such a reconstruction with data from such poorly attested language as Phrygian only adds to the difficulties.
    – czypsu
    Commented Jan 15, 2017 at 11:06
  • 1
    I am far from stating that this etymology is absolutely and surely wrong. I just wanted to present an alternativ, underline the uncertainty of such reconstructions and discourage from presenting them as undeniable facts.
    – czypsu
    Commented Jan 15, 2017 at 11:11

PIE *h2(e)r or *ar- 'thus, so' is the PIE root. Don't think though the Persian word derives from it. Phrygian has ἔρα which is a cognate to the Greek ἆρα.

  • say, does that have to do with either: Latin oralis, os ~ or "mouth, speech", PIE *h₃éh₁os? Or: G reden "to speak, talk", En riddle, and everything belonging to "right", correct? Also cp Ger daher "thus, there from", formally grouped with hier "here", which is neither here nor there.
    – vectory
    Commented Dec 11, 2019 at 13:35
  • This answer is confusing ἆρα, the interrogative particle, with ἄρα, the conjunction, I think. It's true that in poetic language ἆρα was sometimes used as if it were ἄρα but they're distinct.
    – Cairnarvon
    Commented Mar 27, 2021 at 19:24

Warning: I am not a linguist!

This post is essentially a long comment on French’s use of est-ce que and your exclusion of it because “it’s a whole phrase”. I think est-ce que in modern French is not a “whole phrase” anymore, at least in colloquial French, so colloquial French is an example of Indo-European language with such an interrogative particle. (In the citation bellow, Rodney ball mentions the Polish czy, which would be another example).

The difference between formal written French and spoken French are quite large, even in grammar. While est-ce que is currently part of standard French, it doesn’t sound very formal to my native ears and is discouraged in formal writings. I believe it is a relatively recent development in French, but I clearly do not analyse it instinctively as a “whole phrase”. Rodney Ball, in his (excellent) Colloquial French Grammar, section 2.2.1, says

In many ways, est-ce que has come to resemble the interrogative markers which in a number of languages are placed before (or after) statements in order to turn them into questions: czy in Polish, for example. If French lacked a writing system and was being analysed for the first time, est-ce que might well be treated as an indivisible unit: [ɛskə].

This confirms my native instinct. Otherwise, I wouldn’t have needed a linguist to notice the redundancies of my natural way to say “What is this ?” : «Qu’est-ce que c’est», literally, *“what is it that it is”. For me, «Qu’est-ce que» is just a fancy orthography for a word [kɛskə] translating the English “what”.

The same question in 18th century French would have been asked by «Qu’est-ce ?». But est-ce without a following que is no longer used in spoken French, unless you want to sound archaic and/or formal.

Obviously, given the chronology of this development, there is no way est-ce que could be cognate to similar interrogative words in any other language.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.