By "adverbial free relative clause" (maybe there's a better term for this) I mean a relative clause which (a) is headed by an indefinite fused relative pronoun, e.g. English whoever, whatever, and (b) has no antecedent in the main clause and thus no close syntactic connection with it (hence "adverbial"). English examples:

a. Whatever the weather is tonight, we'll go out.

b. Whoever wrote that paper, it's silly.

c. Whichever road you take, you'll get to London.

My questions are:

  1. Does this construction exist in early Indo-European languages (e.g. Greek, Latin, Sanskrit, Hittite)? If not, how do these languages express such meanings?

  2. What syntactic and semantic analyses of such clauses (in any language) have been proposed? They have the unusual property of not having an antecedent and modifying an otherwise syntactically independent sentence, which is odd for relative clauses.

  • I wonder if this might be a derivative of the IE locative absolute (became the well-known ablative absolute in Latin), where the adverbial phrase is a non-finite clause in locative case. Commented Oct 30, 2013 at 16:52
  • 2
    Kyle Rawlins has been working on precisely these constructions, which he calls 'unconditionals', for a few years now. There's a brief description on his website: mind.cog.jhu.edu/~rawlins/research/unconditionals, and his 2008 dissertation on the topic is really interesting: mind.cog.jhu.edu/~rawlins/papers/rawlins_dissertation.pdf. His analysis is quite complex; If i have a little time over the next few days, i'll attempt to summarise it as an answer, but otherwise Rawlins' work is what you should be looking at :-)
    – P Elliott
    Commented Oct 30, 2013 at 21:32
  • 1
    @JustinOlbrantz: I don't think so, because English, Latin, Greek, and many other Indo-European languages have this construction alongside a true absolute, and it is quite different. (The absolute in Greek/Latin/English is not a full clause (has no finite verb), whereas the indefinite adverbial clause has a finite verb in all three languages.)
    – Cerberus
    Commented Oct 31, 2013 at 0:12

4 Answers 4


After a bit of searching I can now partially answer my own question 1, about these constructions in Latin and Greek: apparently both have basically the same construction as English, using an indefinite relative pronoun (usually quisquis in Latin, ὅστις or ὁστισοῦν in Greek) and a verb in the indicative.


hostem qui feriet, mihi erit Carthaginiensis, quisquis erit 'whoever kills an enemy will be a Carthaginian in my eyes, whoever he may be' (Cicero Balbus 22.51)

ita ille faxit Iuppiter, ut ille palam ibidem adsiet, quisquis illest, qui adest a milite 'I wish Jupiter would so make it, that he were openly in the same place with me, whoever he is, that has arrived from the Captain' (Plautus Pseudolus 4.1.22-4)


δεινῶς γ᾽ ἔοικεν ἀποκεκρυμμένην τέχνην ἀνευρεῖν ὁ Τεισίας ἢ ἄλλος ὅστις δή ποτ᾽ ὢν τυγχάνει καὶ ὁπόθεν χαίρει ὀνομαζόμενος 'a wonderfully hidden art it seems to be which Tisias has brought to light, or some other, whoever he may be and whatever country he is proud to call his own!' (Plato Phaedrus 273c)

Greek additionally has a slightly different construction which places the free relative clause inside a conditional:

μὴ ἐπιτρέπειν τῷ ἀσεβοῦντι μηδ᾽ ἂν ὁστισοῦν τυγχάνῃ ὤν '[we ought not] to let him who acts impiously go unpunished, no matter who he may be', lit. 'not even if he happens to be whoever'. (Plato Euthyphro 5e)

  • Good examples, but I see antecedents / close syntactic connections in all of them... (The last example is more literally translated as "not even if he happens to be whoever/anyone").
    – Cerberus
    Commented Oct 31, 2013 at 2:53
  • No, there's only coreference. In an equivalent English example like "The author of that paper is wrong, whoever he is", author is not the antecedent of whoever, right? A fused relative pronoun has no antecedent by definition.
    – TKR
    Commented Oct 31, 2013 at 3:38
  • But there is at least a syntactic connection. It's not the same as the examples in your question.
    – Cerberus
    Commented Oct 31, 2013 at 3:39
  • Coreference is a semantic connection, not a syntactic one, so I'd say they are the same. But your Catullus example is an excellent match; I would have quoted it instead if I'd found it.
    – TKR
    Commented Oct 31, 2013 at 3:48
  • I think I disagree about calling it merely coreference, then. I think it was quite important to Latin authors to always have a noun in the right position and with the same number, sex, and reference present whenever they used an indefinite relative clause. My evidence is that it took me some time to find the example from Catullus (hit number 50 or so), and it is still poetry. And your examples, too, adhere to this rule of the right position, same number etc. It is the same in Greek. The rule is quite different from English and Dutch, and I think that is significant.
    – Cerberus
    Commented Oct 31, 2013 at 4:22

It exists in Greek and Latin, and probably in most other Indo-European languages—at least in the ones I know: Dutch, French, German, Italian.

In order to create an indefinite relative pronoun in Latin, one uses an extended form of the interrogative pronoun, doubling it, so quidquid instead of quid "what"; or one adds -cumque to the relative pronoun, so quodcumque instead of quod "(that) which". The indicative mood is used.

Catullus describes how Egnatius is always grinning, whatever happens, Carmen 39, 5–7:

... si ad pii rogum fili [if at the pyre of a loyal son]
lugetur, orba cum flet unicum mater, [people mourn, when the bereft mother bemoans her only son,]
renidet ille. quidquid est, ubicumquest, [he grins. Whatever it is, wherever he is,]
quodcumque agit, renidet. ... [whatever is happening, he grins]

In Greek, there are many ways to form indefinite pronouns. One can add a form of the interrogative pronoun tis to the relative pronoun hos, but there are other ways. The indefinite clause is normally in the subjunctive mood. An example from Xenophon, Memorabilia 2.3, 6:

ἄλλοις μὲν ἀρέσκειν δύναται, ἐμοὶ δὲ ὅπου ἂν παρῇ πανταχοῦ καὶ ἔργῳ καὶ λόγῳ ζημία μᾶλλον ἢ ὠφέλειά ἐστιν.
[he can be pleasant to others, but to me, whenever he is there, he is in all respects, both in deed and in word, harmful rather than helpful.]
Marchant (1923) translates: "he is pleasant enough to other people, but whenever he is near me, he invariably says and does more to hurt than to help me".

It is much easier to find an antecedentless (and postcedentless) indefinite pronoun when it refers to something adverbial (a time or place) rather than a thing or person. Both Latin and Greek are (far) more likely to add an antecedent than English or Dutch. I was able to find a Latin example, but not a Greek example. Nevertheless, I think the origin of our antecedentless indefinite clause lies in indefinite clauses that do have antecedents, as in Latin and Greek.

An different but interesting development in French is that of the word quoique, "what which", similar to Latin quidquid. It can mean "whatever"; but it has also developed a meaning "however", and even "although", as a full conjunction. That is, there is no longer a "what" to fulfil the function of a subject or object in the subordinate clause. La Harpe to Voltaire (s.xviii):

Quoique éloigné du centre de notre littérature, vous en êtes toujours l'âme et l'honneur. [However far removed from the centre of our literature, you are still/always its soul and honour.] Modifying the adjective éloigné.

Calmet (s.xviii):

Les écrivains sacrés, quoique remplis d'une lumière supérieure et infallible, s'expriment d'ordinaire d'une façon humaine et populaire; ... [The sacred writers, though full of a superior and infallible light, normally expressed themselves in a human and popular manner; ...] Modifying the entire clause (leaving out the finite verb as in English).

Indefinite relative clauses seem to be intimately connected with a concessive ("although") connotation in many languages.

  • The Greek example isn't exactly the type of thing I'm looking for since ὅπου can be analyzed as a temporal conjunction rather than a relative pronoun. The Latin example is nice, though. (Btw "Proto-Indo-European languages" should be "Indo-European languages".)
    – TKR
    Commented Oct 31, 2013 at 4:00
  • @TKR: I suppose you have a point about the Greek example. There is the stem/root of the relative pronoun present in hopou (as in hothen and hote), but they can function exactly like what we call conjunctions. We can say "the time when / the place where this happened", but the absence of an antecedent is perhaps trivial and too common. // Corrected the PIE: I accidentally typed my hotstring pieqq instead of ieqq...
    – Cerberus
    Commented Oct 31, 2013 at 4:29

In PIE such construction would be conducted with a word, consisting of the root a̯ei̯- "always, ever, constant, the same" and a interrogative pronoun, like q̆oteros "which", with a suffix -i to form adverb (formerly, locative case):

a̯i̯q̆oteri - either of the two (the English word either origins exactly this way, via Old English æghwæðer)

a̯i̯q̆ea̯li - of whatever sort, equally


A̯i̯q̆oteri tua̯ pnta̯oi̯ e̯eisi u̯ici bheue̯si. - Either road you go, you will reach the village.

Note that English word "whatever" also produced from the same roots: q̆od "who" + a̯ei̯- "always".

  • What makes you think such a compound existed in PIE and was used in this way? Are reflexes of PIE *h2eiu- used to form indefinite relative pronouns in any IE language other than English?
    – TKR
    Commented Oct 31, 2013 at 2:33
  • @TKR no, it is not specific to English. Compare Latin aequus aequalis. P.S. made a fix.
    – Anixx
    Commented Oct 31, 2013 at 3:05
  • I don't see how aequus can be from *h2eiu- (though aevum is), but in any case it certainly isn't a relative pronoun!
    – TKR
    Commented Oct 31, 2013 at 3:40
  • @TKR the root was a̯ei̯- and the -u- was a suffix. In this case the suffix was not used. For the construction you speak about, the forms with suffix -i were used to make an adverb: a̯i̯q̆ea̯li = equally, a̯i̯q̆oteri = either etc.
    – Anixx
    Commented Oct 31, 2013 at 3:49
  • 1
    Latin aequus isn't a pronoun and isn't used in this type of construction, so even if it was cognate with English -ever (and I don't see how it can be - where would the qu come from?), it wouldn't show that such a construction existed in PIE. You can't reconstruct a feature to PIE based on a single construction in one daughter language (English).
    – TKR
    Commented Oct 31, 2013 at 4:10

You're talking about a circumstantial clause, which refers to a set of circumstances which help explain the rest of the sentence. Circumstantial clauses do appear in Latin, and therefore most of the romance languages (e.g. Spanish, French, Italian, etc.). Although they use relative pronouns, circumstantial clauses aren't relative clauses at all (at least not in Latin).

In Latin, a circumstantial clause always includes the word "cum" (which means "with", as in "magna cum laude" ("with great praise")). "Cum" isn't a relative pronoun at all; it's actually a preposition.

  • The Latin cum construction is completely different from the ones I'm asking about, in both sense and structure. (Btw, this cum is neither a relative pronoun nor a preposition, but a subordinating conjunction.)
    – TKR
    Commented Oct 30, 2013 at 22:49
  • To clarify TKR's comment: this cum is not the preposition cum that means with, but a homonymous word with a different meaning and origin.
    – Colin Fine
    Commented Oct 31, 2013 at 0:22
  • @ColinFine: In the construction TKR and Emsoap are talking about, it is a preposition...
    – Cerberus
    Commented Oct 31, 2013 at 2:58
  • Let's get this straight. There are two 'cum's in Latin. The one used in circumstantial clauses is a conjunction; the one used in prepositional phrases like magna cum laude is a preposition. (I know Cerberus and Colin Fine both know this, just clearing up the confusion in this comment thread.)
    – TKR
    Commented Oct 31, 2013 at 3:43

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.