Although the modern Munda languages show a standard head-final subject–object–verb (SOV) order in unmarked phrases, most scholars believe that proto-Munda was head-first, VO like proto-Austroasiatic.
VO order has been found in compounds, noun incorporation verbal morphology in the Sora[8]-Gorum languages, and to a lesser extent in Gutob, Remo, Kharia, and Juang.
[9][10] Remo: gui-tiwash-handgui-tiwash-hand'wash hand'Sora: ɲen1SGdʒum-te-ti-n-aieat-banana-NPST-INTR-1SG.SUBJɲen dʒum-te-ti-n-ai1SG eat-banana-NPST-INTR-1SG.SUBJ'I am eating banana'Juang: ba-ama-gito-ke1DU.SUBJ-NEG-sing-PRES.TRba-ama-gito-ke1DU.SUBJ-NEG-sing-PRES.TR'We two don't sing'Gorum: ne-r-ab-so’ɟ-om1SG.SUBJ-NEG-CAUS-learn-ACT:2SG.OBJne-r-ab-so’ɟ-om1SG.SUBJ-NEG-CAUS-learn-ACT:2SG.OBJ'I didn’t teach you'Winfred Lehmann (1973) reviewed,[11] "If we examine further evidence provided by Pinnow, we note that Munda contains VO characteristics.
"Pinnow (1963, 1966) proposed that proto-Munda was SVO and that was the syntax of proto-Austroasiatic, which was also highly synthetic like Munda, whereas he attributed analytic and isolating typological features in modern Mon-Khmer to language contact in the Mainland Southeast Asia linguistic area.
[13] They believed that Munda synthesis and SOV order were clearly not stimulated just by language contact within the South Asian linguistic area, but by internal restructuring that caused the Munda word prosody to shift its rhythmic patterns from typical Austroasiatic rising, vowel reduction, iambic stressed to falling, vowel harmony, trochaic stressed profile, thus reversed the clausal syntactic structure and triggered word agglutination.
Stanley Starosta (1967) explained that, during its early formation stage when Munda was still head-first SVO, verb-noun incorporation was facilitated as seen in modern Sora, but then it did a syntactic shift to head-final SOV and added more morphology.
The situation is quite different in South Munda languages, especially Juang, Gtaʔ, and Sora-Gorum, where the original proto-Munda prefix slots are well-preserved, but later additional developments of their predicates are mostly suffixes or enclitics.
[23] Proto-Munda NPs appear to follow proto-Austroasiatic order: [quantifiers] noun [modifiers], whereas modern Munda languages, following South Asian norms, have restructured NP moderately by placing [modifiers] to precede the head noun, while keeping the original proto-Austroasiatic NUM CLF N order like Bahnaric, Vietic, Katuic, Aslian, Nicobarese, and Khasian.
[26] Thus, Munda pronominalized indexation possibly has parallels with some Eastern Austroasiatic languages in the Aslian and Katuic, where prosodically weak resumptive pronouns are used to mark agreement with subjects/agents, but not with objects.