Munda

Origin

Many linguists suggest that the Proto-Munda language probably split from Proto-Austroasiatic somewhere in Indochina. Studies by Chaubey et al. (2011), Arunkumaret al. (2015), Metspalu

et al. (2018), and Tätte et al. (2019) all show that the Munda branch of the Austroasiatic family was created as the result of a male-biased linguistic intrusion into the Indian subcontinent from Southeast Asia during the Late Neolithic period (Sidwell & Rau 2019 cited Tätte et al. (2019), estimate a date of formation between 3,800 and 2,000 years ago), which carried the paternal lineage O1b1a1a into India from either Meghalaya or the sea. These studies and analyses confirm George van Driem's Munda Father tongue hypothesis. Paul Sidwell (2018) suggests they arrived on the coast of modern-day Odisha about 4000–3500 years ago (c. 2000 – c. 1500 BCE) and spread after the Indo-Aryan migration to the region.

Rau and Sidwell (2019), along with Blench (2019), suggest that Pre-Proto-Munda had arrived in the Mahanadi River Delta around 1500 BCE from Southeast Asia via a maritime route, rather than overland. The Munda languages then subsequently spread up the Mahanadi watershed. 2021 studies suggest that Munda languages impacted Eastern Indo-Aryan languages.

Classification

Internal classification

The bipartite Diffloth (1974) classification is widely cited:

Diffloth (2005) retains Koraput (rejected by Anderson, below) but abandons South Munda and places Kharia–Juang with the northern languages:

Anderson's 1999 proposal is as follows.

However, in 2001, Anderson split Juang and Kharia apart from the Juang-Kharia branch and also excluded Gtaʔ from his former Gutob–Remo–Gtaʔ branch. Thus, his 2001 proposal included five branches for South Munda.

Anderson (2001) follows Diffloth (1974) apart from rejecting the validity of Koraput. He proposes instead, on the basis of morphological comparisons, that Proto-South Munda split directly into Diffloth's three daughter groups, Kharia–Juang, Sora–Gorum (Savara), and Gutob–Remo–Gtaʼ (Remo).

His South Munda branch contains the following five branches, but the North Munda branch is the same as those of Diffloth (1974) and Anderson (1999).

Sora–Gorum Juang ↔ Kharia ↔ Gutob–Remo ↔ Gtaʔ

Note: "↔" = shares certain innovative isoglosses (structural, lexical). In Austronesian and Papuan linguistics, this has been called a "linkage" by Malcolm Ross.

Paul Sidwell (2015) considers Munda to consist of 6 coordinate branches, and does not accept South Munda as a unified subgroup.

External classification

Traditionally, the Austroasiatic phylum was divided into two primary groups, Mon-Khmer and Munda, following Schmidt (1906)'s classification. Sebeok (1942) and Pinnow (1959, 1960, 1963) later reinforced this model, attributing the distinction to the typological contrast between the geographically eastern Mon-Khmer (Pinnow's Khmer-Nicobar) and the South Asian Munda branch. This binary classification, however, overlooks lexical evidence, the effects of language contact and areal convergence that may have profoundly shaped the structure of individual branches. Pinnow (1963:150) himself acknowledged that his classification "[...]have only structural and geographical justification." As comparative-historical research progressed, alternative hypotheses concerning the external position of Munda emerged. During the 1970s and 1980s, Diffloth and Shorto independently came up with a hypothesis that suggests recognizing the phylum as made up of two Northern and Southern phylogenetic groups on the basis of shared lexical innovations and the *s-:*t- correspondence in minor syllables. Under this hypothesis, Munda formed part of the Northern group together with Khasian, Palaungic, Khmuic, Mang, and Vietic. Sidwell (2014), however, argues that Munda lacks convincing lexical and structural innovations linking it to this putative Northern subgroup. He further notes that the lexical isoglosses required to unite Mang-Pakanic with the Khasi-Palaungic group are also absent.

A computational phylogenetic analysis study conducted by Sidwell & Greenhill (2014) found no evidence supporting a higher-level Mon-Khmer v. Munda division. Instead, the Austroasiatic family was optimally resolved into eleven primary nodes, with Munda recovered as a coherent coordinate branch equivalent in rank to the other major lineages. The analysis therefore provides no support for treating Munda as the sister group of a monophyletic Mon-Khmer subgroup.

Sociolinguistics and history of Munda studies

The Munda languages are the least documented languages in South Asia. Most Munda languages, except Santali with ~8 million native speakers, can be tentatively described as under threat, with Mundari (~1.5 million), Ho (~1 million), Korku (~400,000), Kharia (~200,000), and Sora (~250,000–300,000) still maintaining substantial everyday usage; smaller South Munda languages such as Juang, Gtaʔ, Remo, Birhor, and Gorum survive in small, fragmented populations amidst more than half a billion Indo-Aryan and Dravidian speakers in one of the most densely populated regions on Earth. Speakers are concentrated in central and eastern India, especially Jharkhand, Odisha, West Bengal, Madhya Pradesh, Maharashtra, and adjacent regions, typically in hill and forest zones, which Peterson (2017) terms accretion zones. Literacy rates are usually low among speakers, and opportunities for daily communication to be written down are largely restricted to the larger languages. Multilingualism is widespread, with frequent use of Hindi, Bengali, Sardi, Odia, Marathi, Telugu, and other regional languages in administration, trade, and education.

Munda languages have long been affected by social stigmas because their speakers, mainly belonging to Adivasi, Dalit, and other socially marginalized communities, have historically occupied subordinate positions within the regional caste and socioeconomic hierarchy. Although folk literatures are abundant, it is not so surprising that the Munda languages lack written literature before the 19th century; the same can be said of other languages spoken by dalit and tribal communities (adivasi) of South Asia (Brahui, Kurukh, Bhili, Burushaski, Nihali, etc). Colonial and postcolonial attitudes often viewed tribal populations as "primitive" and backward, and these perceptions became attached to their languages, which dominant groups frequently regarded as rural and lacking prestige. The dominance of regional Indo-Aryan languages such as Hindi, Bengali, Odia, and Sardi in education, administration, religion, and public life further reinforced this hierarchy, constantly pressuring many Munda speakers to shift toward socially prestigious majority languages.

Long-term intense contact has led to extensive Aryanization. Some communities formerly associated with Munda languages have largely abandoned them and now speak Indo-Aryan varieties instead. Turi is a clear example: although it is genetically a North Munda language, most ethnic Turi no longer speak it fluently with just over 0.56% of the Turi population use the language in daily life. Birhor communities, meanwhile, increasingly adopt Sardi and Hindi. Similar pressures affect Kharia-speaking regions, where prolonged interaction with neighboring Indo-Aryan populations and Kherwarian groups has produced strong lexical and structural convergence. In several South Munda communities, monolingual speakers have disappeared entirely. Certain Indo-Aryan speaking groups in the lowland plain may have originally Munda speakers who went through Aryanization, such as the Cheros of Bihar, the Lohar weavers of Ranchi, the Musahars and Tharus of lowland Nepal, and the Sahariya of Eastern Uttar Pradesh.

Phonology

Consonants, vowels, and syllable structure

The Munda languages share similar sets of phonemes with regional languages in their respective areas. Inherited Austroasiatic "checked" glottalised stop (pre-glottalised articulatory) and nasalised final consonants found in some Munda languages such as Mundari (eg. ub ("hair") is realised as [uˀb̥ᵐ]) and Kharia (eg. oreˀdʒ ("ox") is realised [ɔrɛˀɟ˺ⁿ]) may stand out in South Asia. One key feature in the Munda consonants is the distinction between dental and retroflex stops (/t̪/ vs /ʈ/, /d̪/ vs /ɖ/) in lexical level exists in most Munda languages except Sora and Gorum, reflecting a general characteristic of South Asian (Indosphere) phonology. Because of South Asian areal convergence, Munda languages generally have fewer vowels (between 5 and 10) than their Eastern Austroasiatic relatives. Additionally, Sora has glottalised vowels. Like any other Austroasiatic languages, the Munda languages make extensive uses of diphthongs and triphthongs. Larger vowel sequences can be found, with an extreme example of Santali kɔeaeae meaning ‘he will ask for him’. Most Munda languages lack tones with an exception of Korku, which has acquired two contrastive tones within the South Asian linguistic area: an unmarked high and a marked low, in which the low tone might have emerged from the putative laryngeal feature *χ in a transitive past suffix of Proto-Munda. The general syllable shape is (C)V(C). Initial clusters of the prototypical Austroasiatic sesquisyllables C(C)CVC mostly appear as resyllabified in Munda languages, except Gtaʔ which is the only Munda language that allows initial clusters, eg. gsæŋ "chicken." However, scholars skeptically consider that Gtaʔ sesquisyllabic structure might have been the result of a local secondary development.

As stated above, tonogenesis in Korku and continuous CCVC/sesquisyllabic development in Gtaʔ, both of which were unfolded inside the South Asian linguistic area, which is very hostile to tone and initial clusters, seem to be local innovations and have nothing related to contact-driven restructuring in the subcontinent. It is also unclear whether they had anything connected to areal convergences in the Eastern Austroasiatic languages of Mainland Southeast Asia. Munda word shape is dictated by a general phonotactical phenomenon called bimoraic constraint, which requires free-standing nominal stems to stay disyllabic or to obtain weight at the stressed syllable; that is, monosyllabic free form of lexemes must be expanded (i.e. resyllabified and/or reduplicated) in order to remain heavy (Anderson & Zide 2001). See #Vocabulary for comparison.

Munda phonemes

The following table compiles lists of consonantal and vowel systems of several Munda languages, mainly from Anderson (2014), Nagaraja (1999), and many others on International Phonetic Alphabet.

Word prominence

Donegan & Stampe (2004) posited overarching assumptions that all Munda languages have completely redesigned their word prosodic structure from proto-Austroasiatic rising intonation, iambic and reduced vowel, sesquisyllabic structure to Indic norms of trochaic, falling rhythm, stable or assimilationist consonants and harmonised vowels, making them typological opposed to Eastern Austroasiatic languages at almost every level. Sidwell & Rau (2014) criticised Donegan & Stampe by pointing out that the overall picture appears much more complicated and diverse and that generalisations of Donegan & Stampe are not supported by the instrumental data of the various Munda languages. Peterson (2011b) describes word-rising contour in monosyllables and second syllable prominence in Kharia content words. Even the presence of clitics and affixes does not drive Kharia word prosodic structure to that of a trochaic and falling system. Osada (2008) reports final-syllable stress in all but CVC.CV stems in Mundari. Horo & Sarmah (2015), Horo (2017) and Horo, Sarmah & Anderson (2020) found that the Sora disyllables are always iambic, reduced first syllable vowel space, and second syllable prominence. Even CV.CCə words show final-syllable prominence. Horo & Sarmah (2015) note that the Sora vowels of the first syllables are centralised and that vowels in the second syllables are more representative of the canonical vowel space.

Ghosh (2008) describes about Santali prosody that "stress is always released in the second syllable of the word regardless of whether it is an open or a closed syllable". His analysis was confirmed by Horo & Anderson (2021), whose acoustic data clearly shows that the second syllable in Santali is always the prominent syllable, with a greater intensity of stress and a rising contour.

Zide (2008) reports that in Korku, the final syllable is heavier than the initial syllable, and within a disyllable, stress is preferentially released at the final syllable. The analyses inferred from fieldworks demonstrate that despite exhibiting some variants, most Munda prominence alignments are in line with other Austroasiatic languages, with a predictable final-syllable prominence in a prosodic word. Again, Donegan & Stampe (2004)'s claim on rhythmic holism does not conform with the data presented by individual Munda languages.

Morphology

Morphologically, both North and South Munda subgroups mainly focus on the head or the verb, thus they are primarily head-marking, in contrast to the Indo-European and Dravidian languages, which are mainly dependent-marking. As a result, nominal morphology is less complex than is verbal morphology. Case markers on nominals to show syntactic alignments (nominative-accusative or ergative-absolutive) are largely absent or not systematically developed in the Munda languages, especially the Kherwarian lects. The relation between subject and object in clause is conveyed mainly through verbal referent indexation and word order. At the clause/sentence level, Munda languages are head-final but internally head-first in referent indexation, compounds, and noun incorporation verb complexes.

Munda head-first, bimoraic constraint-free noun incorporation is also found in Khasian, Nicobaric, and other Mon-Khmer languages. In word derivation, besides their own innovative methods, the Munda languages maintain Austroasiatic methods in forms of reduplication, compounding, and derivational infixation and prefixation.

One unusual characteristic that appears to be pervasive among the Munda languages is lexical flexibility, that is, a large number to almost entire their lexicons are precategorial, i.e. lexically underspecified for categories such as noun, verb, adjectives etc. Thus, these languages may rely more on syntax and affixes/clitics to distinguish parts of speech. Pinnow summarised the issue back in the 1960s,

[...]Theoretically any word for any concept, i.e., all words, can function as a verb base. Thus, we may not speak of a verb in the Indo-European sense. This fact was recognized at an early date and is now generally known. This is true for all Munda languages, but cannot be considered distinctive of them since there are innumerable other language families in which the situation is the same.[...] This phenomenon undoubtedly goes back to very ancient times and can probably be accepted as Proto-Munda.

The following elicited examples from Simdega Kharia illustrate the problem:

The degree of lexical flexibility are extremely prominent in North Munda and Kharia, where the lexicons may only contain an open class of contentives; whereas in other South Munda languages this phenomenon however tends to be much weaker.

North Munda

The North Munda subgroup is split between Korku and the 14 Kherwarian languages.

Kherwarian is a large language continuum with speakers extending west to east from the Indian states of Uttar Pradesh to Assam, north to south from Nepal to Odisha. They include fourteen languages: Asuri, Birhor, Bhumij, Koda, Ho, Korwa (Korowa), Mundari, Mahali, Santali, Turi, Agariya, Bijori, Koraku, and Karmali, with the total number of speakers surpassing ten million (2011 census). The Kherwarian languages are often highlighted because their elaborate and complex templatic and pronominalised predicate structures are so pervasive that it is obligatory for the verb to encode tense–aspect–mood, voices, transitivity, and indexation of two arguments, including outside arguments like possessors.

Noun incorporation is often described as an ancestral Munda morphological feature and is essential to the grammar of other South Munda languages such as Sora, but the Kherwarian languages appear to have lost noun incorporation altogether. Nevertheless, rare instances of noun incorporation may be found in some archaic Kherwarian registers and oral literature.

Unlike the Kherwarian languages, with their complex verbal morphology, Korku verbs are moderately simple, with a modest amount of synthesis. Korku lacks person/number indexing of subject(s)/actor (except third persons of locative copulas and nominal predicates in the locative case) and independent present/future tense markers. Korku present/future tenses rely on the finitising suffix -bà. Present or future tense negation can be located in preverbal or postverbal positions, but past tense negation is marked by the suffix -ᶑùn.

Many Korku auxiliary verbs are borrowed from Indo-Aryan. The auxiliary verb takes tense–aspect–mood, voice, and finite suffixes for the lexical verb. An example is ghaʈa-, which means 'to manage to, to find a way to' and serves as the acquisitive.

South Munda

Compared to North Munda languages, South Munda languages are even more divergent and have fewer shared morphological traits. Even the classification of Munda languages is controversial, and South Munda does not seem to exist as a valid taxon. However, South Munda languages retain many notable characteristics of the original Proto-Munda such as prefix slots and scope-ordering of referent indexation and so they represent the less restructured morphology of Munda and reflect the older Proto-Munda and Proto-Austroasiatic structures.

In Kharia, subject markers index not only dual/plural exclusive/inclusive but also honorific status. Objects are not marked in the predicate but instead by the oblique case: -te.

There is a reduplicated free-standing form of finite predicates that behaves differently from the simple predicative base. In the predicate, reduplicated free-standing form never marks tense–aspect–mood and person. That causes the free-standing form to be used in subordination, an attributive function corresponding more or less to relative clauses. Kharia is perhaps the only Munda language that has a infinitive on predicates, =na. The infinitive can serve also as a nominaliser: jib=na=te ‘touching’.

Like in Hindi and Sadani, Kharia has made a calque to form sequential converbs (conjunctive participles) kon (derived from ikon, ‘do’). They denote the completion of an action before another begins.

The negation particle um attaches or fuses person/number/honorific of the subject argument.

Juang exhibits nominative-accusative alignment with unmarked subject/agents and marked objects or patients.

In Juang, a pro-drop language, verbs can index both two core arguments in a transitive predicate, but not frequently. If the arguments are not omitted, referent indexation is largely optional. Juang has a fairly complex tense–aspect–mood system, which is often divided into two sets: I for transitive verbs and II for intransitive verbs. The verb "be" is usually omitted in the present tense and with a predicate adjective in sentences.

Munda lexicon and lexical relation with other Indian language families

Despite some influence from neighbouring languages, the Munda languages generally maintain a solid Austroasiatic and Munda base vocabulary. The most extreme case is Sora, which has zero foreign phonemes. Agricultural-related words from Proto-Austroasiatic are widely shared (Zide & Zide 1976). Words for domesticated animal and plant species like dog, millet, chicken, goat, pig, rice are shared or semantically alternated. There are even specific terms for husked uncooked rice vs cooked rice vs rice (tree), as well as shared words used in rice production and processing like 'mortar', 'pestle', 'paddy', 'sow', 'grind/ground'. The majority of loan words from Indo-Aryan to Munda are quite recent and mostly came from Hindi. The Southern languages like Gutob have received considerable Dradivian lexical influence. A very small number of lexemes seem to be shared between Munda and Tibeto-Burman, probably reflecting earlier contact between the two groups.

It is clear that hundreds of non-Indo-European words in Vedic Sanskrit that Kuiper (1948) attributed to Munda have been rejected through careful analysis. There is a surprising absence of ancient Sanskrit and medieval Indian borrowings of animal and plant names from Munda. Scholars believe that the Munda tribes typically occupied a marginalised and lowly socioeconomic position in the Hinduized society of Vedic South Asia or did not participate in the Hindu caste system and had barely any contacts with Hindus at all. Witzel (1999) and Southworth (2005) proposed that the early non-Indo-European words with prefixes k-, ka-, ku-, cər- in Vedic Sanskrit belonged to a hypothetical 'Para-Munda substratum', which they believed to be part of the Harappan language. That would imply that Austroasiatic speakers might have penetrated as far as the Panjab and Afghanistan in the early 2nd millennium BC. However, Osada (2009) refuted Witzel and considered that those words might have been in fact Dravidian compounds.

Reconstruction

The proto-forms have been reconstructed by Sidwell & Rau (2015: 319, 340–363). Proto-Munda reconstruction has since been revised and improved by Rau (2019).

Writing systems

The following are current used alphabets of Munda languages:

Mundari Bani (Mundari alphabet)

Ol Chiki (Santali alphabet)

Ol Onal (Bhumij alphabet)

Sorang Sompeng (Sora alphabet)

Warang Citi (Ho alphabet)