The correlative diptych (term introduced in Haudry 1973)1 is a specific variety of the adjoined relative construction. The following diagram symbolizes the base form of the construction, with a preposed subordinate clause.
Plain correlative diptych
[ ... [ [ XRel YNom ]C.i ... ]S.subord |
[ ... [ [ ZDem ]C.i ... ]S.main |
- At the highest syntactic level, the sentence is bipartite, consisting of an adjoined subordinate clause and a main clause.
- A certain syntactic category
C
(nominal, adjectival or adverbial) occurs in both clauses, represented by some kind of indefinite or relative pronoun X
in the subordinate clause and some kind of demonstrative or anaphoric pronoun Z
in the main clause. The relation between these two pronouns is called correlation, indicated by the index i
.
- This structure is interpreted as a correlation between the two clauses such that the subordinate clause delimits a certain concept whose core is marked by
X
and which is the semantic base of a certain referent i
which is taken up by Y
in the main clause.
This is the relative construction that prevails in Old Hittite, as in .
. | Hittite | (KBo III Rs. 57f + KUB XXVI 71 6f, ap. Lehmann 1984:179) |
| KASKALz-a | kw-it | assu | utahh-un |
| campaign-ABL | IND-ACC.SG.INAN | booty | bring.home:PST-1.SG |
| With the booty that I had brought home from the campaign, |
| n-at | aped-anda | halissiyan-un |
| CONN-3.INAN.ACC | D3-INST | adorn:PST-1.SG |
| I adorned them. |
The relative pronoun is identical with the indefinite and the interrogative pronoun. The construction is therefore minimally distinct from a paratactic construction translatable as ‘from the campaign I had brought some booty; with that I adorned them’. The pronoun X
, which would not be necessary in such a paratactic construction, identifies the head nominal of the correlative construction.
A typical example from Old Latin is , with correlative pronouns underlined.
. | ab | arbore | abs | terra | pulli | qui | nascentur, |
Latin | [ from | tree:ABL.SG | from | earth:ABL.SG | sprout(M):NOM.PL | REL:NOM.PL.M | be.born:FUT.3.PL ] |
| the sprouts that will rise from the tree from the earth, |
| eos | in | terram | deprimito. | |
| 3.PL.M | in | earth:ACC.SG | push.down:IMP | |
| those you will push back down into the earth | (Cat. agr. 51) |
. | quae | mihi | antea | signa | misisti |
Latin | REL:ACC.PL.N | me:DAT | before | statue(N):ACC.PL | send:PRF:2.SG |
| The statues you sent me the other day, |
| ea | nondum | vidi. |
| it:ACC.PL.N | not:yet | see:PRF:1.SG |
| I have not seen yet. | (Cic. Att. 1, 4, 3) |
shows a slightly advanced degree of grammaticalization, since the relative clause does not introduce a new referent, but evokes one that belongs to the shared experience of speaker and hearer. Moreover, the subordinating function of the relative pronoun is indicated by its front position.
As in non-correlative adjoined constructions, the only word-order change at sentence level is the permutation of the two clauses, producing a postposed relative clause. Apart from the permutation, the only structural change that the inverted correlative diptych displays in contrast with the plain one is that the head nominal stays in the first clause.
Inverted correlative diptych
[ ... [ [ ZDem YNom ]C.i ... ]S.main |
[ ... [ [ XRel ]C.i ... ]S.subord |
from Old Latin illustrates the inverted correlative diptych. In this particular example, Y
is empty.
. | cave | tu | idem | faxis |
Latin | beware | you.NOM.SG | ACC.SG.N:same | do:PRF.SUBJ:2.SG |
| Don't you do the same |
| alii | quod | servi | solent! |
| other:NOM.PL | REL:ACC.SG.N | slave:NOM.PL | use:3.PL |
| which the other slaves tend to do! | (Pl. As. 256) |
Diachronically, this postposed variant of the relative clause is the basis of the postnominal relative clause, which develops from the former by embedding. In fact, embedding of the relative clause in would be marginally possible, and it would thereby cease to be a correlative diptych (and an adjoined subordinate clause).
1 The term ‘co(-)relative’, introduced in Keenan 1985 and taken up, among others, in Culy 1990:234, de Vries
2002:21f and Andrews 2007, has designated different things depending on the author, has led to confusion with ‘correlative’ and is actually not needed.