The difference between prototypical headed CRCs and headless CRCs concerns the absence of the head after the correlative marker (CREL). In headless CRCs the head is not explicitly stated (cf. Andrews 2007:213; Epps 2012:192–193).47 Headless CRCs are illustrated in the following examples from Hindi. In 12) we see a
headless CRC with indefinite reference (reinforced by the use of the particle bhī), whereas 13) has definite reference. In both examples there is no noun after the correlative pronoun jo, as there was in ex. 3) above. 12) Hindi (Montaut 2005:237):48
[jo bhī usne tumse kahā] vah sahī hai
CREL ever 3SG.ERG you.to said that true is ‘Whatever he told you is true.’
13) Hindi (McGregor 1995:51):49
[jo kahtā hū̃] vah sac hai
CREL saying am that true is
‘What I say is true.’
Headless CRCs are attested in the following Munda languages (Table 4): Table 4: Headless CRCs in Munda languages.
Language Source
Sora Bai (1985:188–189), Starosta (1967:242–243) Santali Neukom (2001:200), Ghosh (2008:83)
Kharia Peterson (2011:409)
Gutob Griffiths (2008:644–645, 667–668)
Gtaʔ Anderson (2008b:709)
The first examples to be cited are from Sora, Santali and Kharia, with Gtaʔ and Gutob being discussed afterwards. In Sora, headless CRCs are well attested and are formed with the native marker etente (itente in Bai), which contains the interrogative pronoun ete(n)- ‘what’.50
14) Sora (S Munda – Sora; Anderson and Harrison 2008b:365, ex. 213c):51
dɔ [etente j-ǝn-om-jom-ǝn ɲaɳ-tε-ji] kun batte aninji mεεɳ-tε-ji
so what eat-NMLZ-eat-N.SFX get-NPST-3PL that with they live-NPST-3PL ‘so whatever food they get that’s what they live on’
In Santali, headless CRCs are attested with the native indefinite pronouns jãhãe ‘whoever’ (animate) and
jãhã ‘whatever’ (inanimate).52
47 Another widely used term for headless CRCs is free (headless) relatives. However, this is imprecise, as headless CRCs are only a subtype of free relatives. For instance, free relatives do not necessarily contain the co-referent demonstrative in the main clause, as in [Whoever comes] will be welcome. They can also be postnominal, as in You
can invite [whoever you want].
48 Glosses have been slightly modified. 49 Glosses are ours.
50 This fact is nowhere stated as such, but can be inferred from various places in Anderson and Harrison (2008b). There the form etente is either glossed as ‘what’ (e.g. on p. 325, ex. 79b or on p. 345, ex. 140a) or analyzed as ete-n-
te, where ete- is ‘what’, -n- an unidentified nominal suffix and -te a focus marker (e.g. on p. 365, ex. 213c).
51 Glosses have been slightly modified. The example is originally from Starosta (1967:243, ex. 4).
52 It appears that the indefinite pronoun spread from such contexts to the prototypical headed CRCs (see fn. Error!
15) Santali (N Munda – Kherwarian; Neukom 2001:200):
[nukin jãhãe-ge-kin hɔrɔk'-a-e]
these(ANIM).DU whoever-FOC-3DU.SUBJ put-APPL-3SG.OBJ
uni-ge raj-e hoe-y-ok'-a
that(ANIM)-FOC king-3SG.SUBJ become-y-MID-FIN ‘Whomever they put it (i.e. a chain) on, he shall be king.’
In Kharia, the headless CRCs exemplified in Peterson (2011:409) contain native interrogative pronouns. 16) Kharia (S Munda – Kharia; Peterson 2011:421–422):
[a=kaɽ seŋ ɖaʔ kuy=e], ho=kaɽ u
Q=SG.HUM first water find=ACT.IRR that=SG.HUM this
daru=te=ga yo=ga ɖe=na”.
tree=OBL=FOC see=FOC come=MID.IRR
‘He who first finds water, he should come to this tree, looking [for the others].’
It may be recalled that in Gtaʔ and Gutob, headless CRCs are the only instances of CRCs attested. Headless CRCs with indefinite reference are illustrated in the following example from Gtaʔ:
17) Gtaʔ (S Munda – Gtaʔ, Anderson 2008b:709):
[ja par-le] mœ paɳ ccoɳ diɳ-le
CREL can-OPT he come REDPL:eat AUX-OPT ‘Whoever wins, let him come and eat.’
The form ja, glossed here as CREL is in fact the interrogative animate pronoun (Anderson 2008b:707). One can assume it is unrelated to the Indo-Aryan (IA) correlative pronoun, given the fact that an identical form is found in Remo (Anderson and Harrison 2008a:579).53 Headless CRCs with definite reference are illustrated
with the following example from Gutob, where the native interrogative pronoun laj is employed:54
18) Gutob (S Munda – Gutob-Remo, Griffiths 2008:668):
[laj mara + mari ɖeɳ-gu-men] o-maj razi ɖem-to
CREL beat + ECHO become-MID.PST-PL OBJ-3.PL agreed make-HAB ‘He makes those who have fought with each other settle their dispute.’
A separate discussion of headless CRCs was warranted mostly because there appears to be evidence suggesting that this construction is original in Munda, and not borrowed from IA.55 This claim is supported
by the fact that headless CRCs are found in numerous languages where there are no headed CRCs, including Dravidian and Tibeto-Burman (see fn. Error! Bookmark not defined. above), as well as in languages from other parts of the world.56 This shows that headless CRCs are more common than headed CRCs, which are
restricted to a very small number of languages.57
53 Or it is borrowed in both languages from IA, as suggested by one of the reviewers.
54 Griffiths (2008:667) claims that either a native or borrowed item can be utilized as the correlative pronoun in Gutob. No example is given of the construction with the borrowed item, and Judith Voß (p.c.) informs me that she has found no instances of the IA correlative pronoun in her corpus.
55 Of course, the IA origin is still possible. Headless CRCs are well attested in IA languages, e.g. in Marathi (Pandharipande 1997:86-87) and Hindi – in addition to the examples cited as ex. 12) and 13) above, there are numerous instances of headless CRCs found, e.g. in McGregor (1995:47ff., 91ff.), Jain (1995:312–313), and Montaut (2004:235–238).
56 This can be inferred from numerous examples cited by Lipták (2004) for Hungarian, Rebuschi (2009a, 2009b) for Basque and Cable (2009) for Tibetan.
57 The headed CRCs are crosslinguistically rare, in particular as a major or main relativization strategy (this was already pointed out in §1).
A claim essentially identical to ours here, namely that headless CRCs should not be considered a borrowing from IA, is promoted for Dravidian by Bai (1985). The author claims that headless CRCs are a native Dravidian construction, which is based on multiple arguments, including the attestation of headless CRCs in the earliest attested Dravidian texts, as well as in various tribal Dravidian languages. Bai briefly discusses Munda languages as well, specifically headless CRCs in the tribal language Sora, extending her claim so as to suggest the native origin of headless CRCs in Sora. The native origin of Munda headless CRCs seems to be further supported by the observation that all the examples of that construction in Munda we have encountered so far contain only native interrogative pronouns (see above). Interestingly, according to the examples from Subbārāo (2012a, 2012b), the same is true for Dravidian headless CRCs.58
The native origin of Munda headless CRCs could help us reconstruct the manner in which borrowing of CRCs into Munda languages played out, as the prior existence of headless CRCs in Munda may have facilitated the borrowing of the IA headed CRCs into Munda by serving as a so-called “common pivot” (Matras and Sakel 2007).