Correlative Clauses

Book Nav

267.1 We have first to distinguish between the simple structure in which the relative clause only qualifies a noun or pronoun in the principal clause, as

τῶν οἳ νῦν βροτοί εἰσι
of those who are now living

ἐν πεδίῳ ὅθι περ κτλ.
in the plain where, etc.

and the parallel structure, in which the relative is an adverb of the same form as the antecedent.

τὼς δέ σʼ ἀπεχθήρω ὡς νῦν ἔκπαγλʼ ἐφίλησα

τόφρα δʼ ἐπὶ Τρώεσσι τίθει κράτος, ὄφρʼ ἂν Ἀχαιοὶ κτλ.

τῇ ἴμεν ᾗ κεν δὴ σύ, κελαινεφές, ἡγεμονεύῃς

Here the notion given by the adverbial ending—manner, time, way, etc.—is the point of comparison, and must be understood to qualify both classes. In both these kinds of compound sentence the demonstrative antecedent may often be omitted, but this is especially the case with the second, in which a relatival adverb implies a corresponding demonstrative. Thus ὡς ἐφίλησα implies τὼς  . . . ὡς ἐφίλησα; ὄφρʼ ἄν is equivalent to τόφρα . . . ὄφρʼ ἄν, etc.

In this way, then, it came about that ὡς (lit. in which manner) means in the manner in which, and so

ὄφρα
to the time up to which


by the way which

ὅθι
at the place where

ὅτε
at the time when

and so on.1 The whole relative clause in fact serves as an adverb (of manner, time, way, etc., as the ending may determine), construed with the verb of the principal clause. Such clauses accordingly are called adνerbial while clauses which merely qualify a noun or pronoun are adjectival.

2. The omission of the antecedent from the governing clause leads to various idiomatic uses.

a. The relative clause comes to be equivalent to a noun or pronoun in any case which the governing clause may require.

ll. 5.481 τά τ ἐέλδεται ὅς κʼ ἐπιδευής
              which (he) desires who is in need

Il. 1.230 δῶρʼ ἀποαιρεῖσθαι ὅς τις σέθεν ἀντίον εἴπῃ
              to take away gifts (from him, from anyone) who, etc.

Il. 7.401 γνωτὸν δὲ καὶ ὃς μάλα νήπιός ἐστιν

Od. 15.281 αὐτὰρ κεῖθι φιλήσεαι οἷά κʼ ἔχωμεν
                   you will be entertained (with such things) as we haνe

Il. 14.81 βέλτερον ὃς φεύγων προφύγῃ κακόν
              it is better (for one) who by flying escapes evil
              i.e. it is better when a man, etc.

Cp. Od. 15.72, Il. 3.109.

b. The omission is especially characteristic of clauses with ὅτε when (for τὸ . . . ὅτε )

Il. 15.18 ἦ οὐ μέμνῃ ὅτε
              do you not remember (the time) when

Il. 8.229 πῇ ἔβαν εὐχωλαί, ὅτε δὴ κτλ.
              where are gone the boastings (of the time) when, etc.

Il. 19.337 λυγρὴν ἀγγελίην ὅτʼ ἀποφθιμένοιο πύθηται

and with numerals.

Il. 21.80           ἠὼς δέ μοί ἐστιν
              ἥδε δυωδεκάτη ὅτε κτλ.
              this is the twelfth morning (from the time) when, etc.

So in Il. 2.303 χθιζά τε καὶ πρωΐζʼ ὅτε means a day or two (frοm the time) that. Hence too the forms εἰς ὅτε to the time that, πρίν γʼ ὅτε before the time when.

Similarly with ὅθι where, as ἵκανον ὅθι they came (to the place) where.

c. With a verb of saying or knowing the relative clause has apparently the force of a dependent question.

ll. 2.365 γνώσῃ ἔπειθʼ ὅς θʼ ἡγεμόνων κακός, ὅς τέ νυ λαῶν,
              ἡδʼ ὅς κʼ ἐσθλὸς ἔῃσι
              you will recognize (γιγνώσκω, not οἶδα) of the leaders
              him who is a weakling, and who of the people,
              and again him who shall be
(found to be) braνe

So Il. 13.278, 21.609, Od. 3.185, 17.363; compare the form with the antecedent expressed.

Il. 23.498           τότε δὲ γνώσεσθε ἕκαστος
               ἵππους Ἀργείων, οἳ δεύτεροι οἴ τε πάροιθεν

The construction is the same with a verb which implies knowing, finding out, or the like.

κλήρῳ νῦν πεπάλασθε διαμπερὲς ὅς κε λάχῃσι
cast lots (tο find him) whose portion it shall be

3. The suppressed antecedent, again, may have no clear or grammatical construction

a. This is especially found when the relative clause expresses reason.

Od. 4.611 αἵματός εἰς ἀγαθοῖο, φίλον τέκος, οἷ’ ἀγορεύεις
                you are of good blood (seeing the things) such as
                you speak

                i.e. as I see by the manner of things that you speak.

Il. 14.95 νῦν δέ σευ ὠνοσάμην πάγχυ φρένας οἶον ἔειπες
              I blame your thought, because of the kind of thing
              yοu haνe said

Od. 2.239 νῦν δʼ ἄλλῳ δήμῳ νεμεσίζομαι, οἷον ἅπαντες
                 ἧσθʼ ἄνεῳ
                 at the way that you all sit silent

ll. 17.586 Ἕκτωρ, τίς κέ σʼ ἔτʼ ἄλλος Ἀχαιῶν ταρβήσειεν,
                 οἷον δὴ Μενέλαον ὑπέτρεσας;
                 who would fear you any more,
                 seeing the way you shrank before Menelaus?

Od. 15.212 οἷος ἐκείνου θυμὸς ὑπέρβιος, οὔ σε μεθήσει

Il. 16.17 ἠὲ σύ γʼ Ἀργείων ὀλοφύρεαι ὡς ὀλέκονται

Od. 10.326 θαῦμά μʼ ἔχει ὡς κτλ.
                   I wonder at the way that, etc.

This is the idiom generally described by saying that οἷος is put for ὅτι τοιοῦτος, ὡς for ὅτι οὕτως, and so on. So when ὅς introduces a reason (§ 266) we might say that it is for ὅτι οὕτος (e. g. Ζεὺς αἴτιος ὅς τε δίδωσι = ὅτι οὕτος δίδωσι). The peculiarity, however, of the clauses now in question is that the relative can have no grammatical antecedent, that is to say, that the correlative which it implies as an antecedent has no regular construction in the principal clause.

b. This is also found after verbs of knowing, etc.—the relative clause expressing the object or thing known.

Il. 2.409 ᾔδεε γὰρ κατὰ θυμὸν ἀδελφέον ὡς ἐπονεῖτο
              he knew his brother (as to the manner) in which he labored

Il. 24.418 θηοῖό κεν . . .
                οἷον ἐερσήεις κεῖται

Od. 7.327 εἰδήσεις . . . ὅσσον
                 ἄρισται νῆες ἐμαί

This is evidently an extension of the form γνώσῃ ὃς κακός (see 2.c, above), with the difference that the suppressed correlative in the principal clause is without a regular construction.

c. Sometimes the relative clause is used without any principal clause, as an exclamation.

Il. 7.455 ὢ πόποι, Ἐννοσίγαι’ εὐρυσθενές, οἷον ἔειπες

Od. 1.32 ὢ πόποι, οἷον δή νυ θεοὺς βροτοὶ αἰτιόωνται

ll. 5.601 ὢ φίλοι, οἷον δὴ θαυμάζομεν Ἓκτορα

The ellipse gives an expression of surprise: (to think) what a thing you haνe said! (to see) hοw men blame the gods! (to remember) how we wondered at Hector! The want of a construction has much the same effect as with the exclamatory use of the nominative (§ 163). Similarly

Od. 4.240 πάντα μὲν οὐκ ἂν ἐγὼ μυθήσομαι οὐδʼ ὀνομήνω,
                 ὅσσοι Ὀδυσσῆος ταλασίφρονός εἰσιν ἄεθλοι·
                 ἀλλʼ οἷον τόδʼ ἔρεξε κτλ.
                 I will not tell all of his feats, but (just to mention)
                 what a feat this was that he did, etc.

So Od. 4.269, 11.517; cp. also Il. 5.638 ἀλλʼ οἷόν τινά φασι κτλ. (just to instance) the kind of man that they tell, etc.

If the explanation now given of these relative clauses is right, it is evidently incorrect to accent and punctuate as is done by editors (e.g.) in

Il. 6.108 φὰν δέ τινʼ ἀθανάτων ἐξ οὐρανοῦ ἀστερόεντος
              Τρωσὶν ἀλεξήσοντα κατελθέμεν· ὣς [or ὧς] ἐλέλιχθεν

taking it as an independent clause "so they wheeled." The same editors do not hesitate to write in Il. 16.17 ὀλοφύρεαι, ὡς ὀλέκονται, where the construction ie precisely the same.

It is sometimes maintained that in all such cases we have a survival of the primitive parataxis—that (e. g.) ὀλοφύρεαι ὡς ὀλέκονται was originally

ὀλοφύρεαι, ὧς ὀλέκονται you lament, they so perish

hence you lament how they perish, or that they thus perish. On the same view the exclamatory οἷον ἔειπες is not elliptical, but represents the original independent what a thing you have said! (See Mr. Leaf on Il. 2.320 θαυμάζομεν οἷον ἐτύχθη). This hypothesis, however, is not borne out by the facts of language. In the first place, it is strange that the traces of parataxis should be found with the relatives ὡς, οἷος, ὅσος, etc., rather than with the corresponding demonstrative forms. Again, if the relative retained an original demonstrative use, we should expect to find this, like other survivals, in some isolated group of uses, whereas the clauses now in question are very various in character. Again, the passages which favor the notion of parataxis are indistinguishable in structure from others to which it cannot be applied, such as most of the examples given under 2. Yet we cannot separate τά τʼ ἐέλδεται ὅς κʼ ἐπιδευής from φιλήσεαι οἷά κʼ ἕχωμεν, or that again from ὠνοσάμην οἷον ἔειπες. In particular it will be found that the theory does not apply to clauses which are conditional so well as to those which give a reason. The exclamatory use—οἷον ἔειπες and the like—does not furnish a good argument, because the pronoun used in a simple exclamation would not be demonstrative, but interrogative (ποῖον ἔειπες, etc.). The most decisive consideration, however, is that the relatival use of ὅς and its derivatives is common to Greek and Sanskrit, and may be regarded therefore as Indο-European. Consequently there is a strong presumption against any hypothesis which explains the Homeric use of the relative from a still earlier or pre-Indo-European stage of language.

4. Sometimes an antecedent is not construed with the governing clause, but follows the case of the relative. This is allοwed if the antecedent is separated from its own clause.

Il. 14.75 νῆες ὅσαι πρῶται εἰρύαται ἄγχι θαλάσσης
              ἕλκωμεν

So ll. 6. 39θ, 10- 415, 14. 371. This "inverse attraction" may be placed with the forms in which the antecedent is wanting, because it can only arise when the original construction of the antecedent (ἕλκωμεν νῆας ὅσαι—) has been forgotten.

5. Again, the correlative structure is liable to an extension, the characteristic of which is that the relatival adverb has no proper construction in its own clause.

This may be most clearly seen in the use of οὅνεκα (i.e. οὗ ἕνεκα) for which reason.

ll. 1.110 ὡς δἡ τοῦδʼ ἕνεκά σφιν ἑκηβόλος ἄλγεα τεύχει,
             οὕνεκʼ ἐγὼ . . .
             οὐκ ἔθελον κτλ.
             Apollo causes sorrow for this reason, that I would not, etc.

Here we cannot translate οὕνεκα for which reason; the reason does not precede, but is given by the relative clause. That is, the first ἕνεκα is rational; the second is logically unmeaning. Hence the οὕνεκα can only be due to the correlation; as it is usually expressed, οὕνεκα is attracted to the antecedent τοὔνεκα. Then—since οὕνεκα comes to imply a correlative τοὔνεκα—the antecedent τοὔνεκα is omitted, and the relatival οὕνεκα by itself comes to mean for the reason that, because.

The process may be traced more or less distinctly in all the relatival adverbs. Thus ὡς (in which manner) comes to mean in such manner that and so ὄφρα for so long that, νᾷ (lit. where) to the end that. Also, as will be shown presently, ὅ, ὅτι and ὅ τε are adverbial accusatives, meaning literally in which respect, hence in respect that, because; cp. εἰπεῖν ὅ τι ἐχώσατο to say for what he was angered with χώσατο ὅτι he was angered for (the reason) that. The qualifying force of the adverb is transferred from its own clause to the verb of the governing clause.

On the same principle ἐκ τοῦ ὅτε from the time when becomes ἐξ οὅ (for ἐκ τοῦ οὗ—) and εἰς τὸ ὅτε becomes εἰς ὅ to the time that.

  • 1. In the corresponding sentences in English it is often the relative that is wanting: thus τῇ ἴμεν ᾗ κεν ἡγεμονεύῃς to go by the way [by which] you lead. This forms a characteristic difference between Greek and English syntax.