Akabea belongs to the Great Andamanese family, one of the two indigenous language families of the Andaman Islands. All that remains today of this family is a handful of “rememberers” of so-called Present-day Great Andamanese, based on the language furthest removed from Akabea. None of the traditional languages was the subject of professional linguistic documentation or analysis. However, two government officials collected extensive material, with Akabea being the language most thoroughly treated. Neither was a trained linguist, and one might wonder whether anything reliable can be derived from their documentation. We have worked in detail with the Akabea material, and conclude that while there are obvious gaps (e.g. in the phonetics), the overall picture is that of a very consistent and elaborate grammatical and lexical structure. We present two typologically unusual features of Akabea. One relates to the conceptual basis of the language’s lexicon, which makes extensive use of somatic (body-part) prefixes, e.g. Akabea aka- ‘mouth’. These are sometimes used literally (as in the word for ‘mouth’), sometimes by semantic extension (e.g. in words relating to food and speech). The other concerns an unusual grammatical feature of the language, namely verb root ellipsis. In English, it is possible to omit verbs under appropriate pragmatic conditions, e.g. in response to Is he reading? one can answer Yes, he is, with omission of the whole word reading, but not *Yes, he is -ing, with omission of just the root read. Precisely this bizarre non-existent English type is what is attested in Akabea.


The part of the world that we will be dealing with in this article is the Andaman Islands, an archipelago located in the Indian Ocean, more precisely in the Bay of Bengal. Today, the Andaman Islands together with the neighboring Nicobar Islands constitute India’s Union Territory of the Andaman and Nicobar Islands.

There are at least two indigenous language families spoken on the Andaman Islands: Great Andamanese, the language family that we will be concerned with here, and Ongan. Linguistically they are quite distinct from one another – this was recognized even by non-linguist administrators in the Andaman Islands in the nineteenth century – and they also represent quite distinct cultures and are quite distinct in terms of population genetics (BARIK et al., 2008). Figure 1 shows the location of the languages in the nineteenth century.

Figure 1.Figure 1. Languages of the Andaman Islands.Source (slightly modified):

The Ongan family has two known languages in it, Önge and Jarawa. Önge is spoken on Little Andaman Island, the southernmost of the major islands that constitute the archipelago, while Jarawa is spoken not only in parts of South Andaman Island, the southernmost of the three large islands that constitute Great Andaman, as in the nineteenth century, but today also further north in parts of Middle Andaman Island. In addition, on North Sentinel Island the language which we call Sentinelese is spoken. This population is virtually uncontacted; we know very little about them and virtually nothing about their language, though their material culture is more similar to that of Ongan-speaking groups, so their language may well be a third member of the Ongan family.


Ongan family



The language family with which we will be concerned here is Great Andamanese. Traditionally, ten distinct ethno-linguistic groups have been recognized within Great Andamanese, although for some of these varieties we have virtually no documentation, indeed for one (Akabo) basically all we have is a few lexical items none of which is distinct from those of the neighboring varieties. So although there are traditionally considered to be ten varieties there may well not have been as many languages in the sense of mutually unintelligible speech varieties, perhaps as few as six, as we have indicated in (2). For justification of the subgrouping, see Comrie & Zamponi (2019a).


Great Andamanese family

South Andamanese:



Middle Andamanese:



[Okojuwoi, Okol]

North Andamanese-Akakede:


North Andamanese

[Akabo, Akachari; Akajeru, Akakhora]

Going roughly from south to north we have the two South Andamanese languages, Akabea – the language that we will be examining in more detail below – and Akarbale. Then come the three varieties grouped together as Middle Andamanese: Okojuwai, Okol, and Opuchikwar; Okojuwai and Okol are very close and may even have been a single dialect chain. Then in the north we have North Andamanese-Akakede, i.e. Akakede plus the four North Andamanese varieties, which may well have been four dialects of a single language with high levels of mutual intelligibility.

The Ongan languages are still spoken today by small but viable speech communities, though of course they are vulnerable because of their small size. The traditional Great Andamanese languages, however, are all extinct. What survives of this family, what has come to be called Present-day Great Andamanese, is an amalgam of North Andamanese varieties based mainly on Akajeru. As of April 2020, there are three semi-fluent speakers, or perhaps more accurately remembers, of the language. They do not usually speak the language, the community language being Hindi. There has been substantial recent work on Present-day Great Andamanese by Anvita Abbi: a dictionary (ABBI, 2012) and a grammar (ABBI, 2013), as well as a number of articles; this is the first extensive work on any Great Andamanese language meeting current standards of linguistic description. Otherwise, all Great Andamanese languages are extinct; the last ethnic Akabea/speaker of Akabea died sometime between the 1921 and 1931 censuses – this was not a case of language shift, the last speaker of the language was also the last member of the ethnic group.

The indigenous inhabitants of the Andamans were largely isolated from the outside world through to the mid-nineteenth century, though this was certainly not complete isolation. For instance, we know that before European contact they had acquired pottery, probably through contact with an Austroasiatic-speaking group, and outrigger canoes, probably through contact with an Austronesian-speaking group (Zamponi & Comrie 2020: 32-35 and references cited there). But this state of near-isolation changed abruptly in 1858 when the British established a permanent settlement and penal colony at Port Blair in traditional Akabea territory. They brought over prisoners from the South Asian mainland who were considered particularly dangerous, including political prisoners. Since the islands are isolated it was a relatively open penal colony, so many of the prisoners actually had their own houses, their own kitchen gardens, etc. It was assumed they would not escape because the indigenous people would be hostile to them; and if they did manage to get a boat and sail away they were in very dangerous waters, so the likelihood of anyone escaping was minimal. However, this did lead to a rapid demographic decline among the Great Andamanese populations, largely due to imported diseases to which the indigenous people had no immunity. So in the years immediately after 1858 there were at least four major epidemics that swept across the Great Andamanese populations – though not the speakers of Ongan languages, who lived more isolated lives – epidemics of pneumonia, syphilis, measles, and influenza.

Given that the traditional Great Andamanese languages are no longer spoken and that Akabea has not been spoken for almost a hundred years, one might wonder what the documentation of these languages is like, in particular the documentation of Akabea. The documentation of Akabea is all from the late nineteenth century into the early twentieth century, with most of it from two colonial administrators. Maurice Vidal Portman worked closely with native speakers of Akabea. He published a manual of the Andamanese languages (PORTMAN, 1887) which consists of a dictionary of about 1,000 entries and a phrase book – what he called “Dialogues” – designed for his successors as colonial administrators. Portman also produced some other material including three short narrative texts, in total about 148 words, which is all the natural narrative material we have for the Akabea language, and a second dictionary of over 2,000 entries (PORTMAN, 1898). Another colonial administrator, Edward Horace Man, actually worked largely with non-native speakers. It seems that in the early years of the penal colony Akabea acquired some status as a local lingua franca among the indigenous populations, so he worked mainly with non-native speakers around Port Blair. He wrote an unpublished grammar which is in the archives of the Royal Anthropological Institute in London, and also a more extensive dictionary of about 3,500 entries (MAN, 1919–23). In addition, one of the founders of British social anthropology, indeed modern social anthropology overall, Alfred Reginald Radcliffe-Brown, worked on North Andamanese but also provided some other linguistic material, including some from Akabea. It should be emphasized that none of these was a linguist, and certainly not a phonetician. So given this perhaps unpromising sounding material, what are we able to say about the language? Well, as they say, “the proof of the pudding is in the eating”, so in order to see in detail what we have been able to extract from the material we refer you to our detailed study Zamponi & Comrie (2020), hereafter abbreviated as “Z&C”. The remainder of this article is a distillation of a few of the more salient points from this larger work, which constitutes the first modern analysis of the grammar of a traditional Great Andamanese language.

1. Limits on our knowledge of Akabea

Given the limitations of the documentation, there are clear limits on our knowledge in several areas. For instance, in phonology, we know from the discussion provided by Portman and Man that the Great Andamanese languages, including Akabea, have retroflex consonants as distinct phonemes. But they then go on to say that since they did not hear the distinction consistently they decided not to try to represent it. So we know that there were distinct retroflex consonants, but we do not know exactly where they occurred. We can get some information by comparing with Present-day Great Andamanese, thanks to the work of Anvita Abbi. Present-day Great Andamanese clearly has phonemically distinct retroflex consonants [ʈ], [ʈʰ], [ɖ], and [ɽ]. Where there are cognates it would be a reasonable hypothesis, though of course not fully guaranteed, that there would be retroflex consonants in Akabea; for many lexical items, there are no cognates.

We can say very little about the discourse structure of Akabea. We only have six spontaneous texts ­– note that Portman’s dialogues are not spontaneous texts, rather he constructed them and elicited Akabea versions. The six spontaneous texts comprise the three narratives already alluded to, and three songs which provide problems of their own: even to native speakers at the time not everything in the songs was intelligible.

2. Intricate complexity

However, there are other areas where the available material gives an insight into the intricate grammatical complexity that characterizes the traditional varieties of Great Andamanese languages. For instance Akabea has twelve series of personal pronouns, not matched directly by anything in the morphology of full noun phrases, of which five express various oblique adverbial relations – benefactive (‘for’), ablative (‘from’), comitative (‘together with’), versative (‘towards’), and interessive (‘among’) – and seven express core relations like subject, direct object, indirect object. The latter includes four used only for subjects plus an additional one used for subjects and some other core arguments. There is thus a rich set of distinctions made in the marking of subjects in the pronoun system.

Series I, for instance, is used for an emphasized subject, as in (3).


wai d-ol ab-pail yaba=da


I am not a woman’ (Z&C, p. 169)

As illustrated by the series I form d-ol, each of these pronoun forms consists of a prefix which indicates the person and number, or sometimes just person, followed by a base which indicates the series.

If the subject is not emphasized and the tense is non-past, then, with the exception of certain sentence types to which we return momentarily, one uses type II, as in (4), where there is no emphasis on the subject.


d-o mami-ke

1SG-II sleep-nPST

‘I sleep, I will sleep’ (Z&C, p. 170)

But if this were in a past tense or if it were in a dubitative polarity question or a relative clause, then one would use series III, so that we have first person singular d-a in the past tense main clause in (5), and second person singular ŋ-a in the relative clause contained within example (6).


d-a mami-re

1SG-III sleep-PRET

‘I slept’ (Z&C, p. 170)


ka miʧiba putu l-Ø-ot-teŋ=da ŋ-a kop=yate

this what wood DEF=3-SP-name=COP 2-III cut=REL

‘What is the name of this wood that you are cutting?’ (Z&C, p. 171)

Example (6) shows that series III must be used in a relative clause even if the time reference is non-past.

Series IV is restricted to subjects of nominalizations. Nominalizations are frequent in Akabea, as in some other Great Andamanese languages, because the basic way of negating a verbal clause like ‘I know the way’ is to nominalize the clause and then make that nominalization the subject of the verb ‘to be’, i.e. something like ‘my knowing the way is not’ or ‘there is not my knowing the way’. In this construction and in nominalizations more generally a subject pronoun from series IV is used, as in (7).


d-ona tiŋa gad-ŋa yaba=da

1SG-IV road know-NMLZ NEG=COP

‘I do not know (the) way’ (Z&C, p. 171)

Incidentally, series IV pronouns are specific to nominalizations. They are not general possessives, as one might expect given that in many other languages the subject of a nominalization is treated like a possessor.

Finally, for subjects but also for other core arguments (direct objects and indirect objects) there is a fifth type, series VII, which consists just of a consonant which is procliticized to the verb. This series can only occur with a verb form that begins with a vowel. This is illustrated in (8) for a subject, in (9) for an indirect object.


wai d=ar-lu-ke

FOC 1SG.VII=SP-finish-nPST

‘I will finish (it)’ (Z&C, p. 173)




‘Show me (that)!’ (Z&C, p. 174)

So what we see here is a very intricate system for encoding subjects. The material is not quite 100% consistent, though it is maintained by and large systematically across all the documentation, and the few exceptions may well reflect misunderstandings or other errors.

The Akabea subject-marking system is summarized in (10).


I emphasized subject

except in dubitative polarity questions, relative clauses, nominalizations

II non-emphasized subject in non-past

except in dubitative polarity questions, relative clauses, nominalizations

III non-emphasized subject in past, dubitative polarity questions, relative clauses

IV subject of nominalization

VII non-emphasized core argument before vowel only2

i.e. subject, direct object, indirect object

3. Somatic (body-part) prefixes

In sections 3 and 4 we will look at two areas in somewhat more detail to illustrate two typologically interesting features of Akabea. For the most part this discussion would also hold for other Great Andamanese languages. The first phenomenon we will present, somatic (or body-part) prefixes (SP), is certainly true for all Great Andamanese languages.

These are prefixes that refer to body parts like head or back or extremities (hands and legs). They are a major feature of the language. In fact, in Zamponi & Comrie (2020), some 57 pages (87-134) of the roughly 250 pages devoted to the grammar itself treat somatic prefixes. Of the 148 words in Portman’s narratives, 24, or roughly one in six have a somatic prefix. The prefix can attach to noun roots, to verb roots, to adjective roots, and to adverb roots. Crucially these prefixes, although they refer to a body part, are not cognate with the corresponding noun. Thus the prefix for ‘head’ is not cognate with the noun root for ‘head’, etc. Indeed some of the prefixes have a semantic range for which there is no corresponding noun. One of the prefixes, ar-, covers the abdomen, the back, and the legs; there is no single lexical item which covers that range. There are eight somatic prefixes in Akabea, as set out in (11), although two of them are rare and may have been going out of use in the system; but the other six are very frequent. In the main part of this section, we will concentrate on the somatic prefix aka- ‘mouth’.


ab-, a- ‘body’

aka- ‘mouth’

ar- ‘abdomen, back, legs’

aya- ‘(inside of) ear’ (rare)

ig-, iʤ- ‘face, arms’

on-, ɔyo- ‘extremities (hands, feet)’

ot-, ɔt- ‘head’

ɔkɔ-, ɔko- ‘lips’ (rare)

Somatic affixes are rare typologically. Figure 2 shows the other language families of the world in which we are aware of their existence, all of them located in the Americas. While there may well be areal diffusion in the Pacific Northwest of North America, and separately in Mesoamerica, otherwise these, including Great Andamanese, seem to be independent instances of a cross-linguistically rare phenomenon.

Figure 2.Figure 2. Language families with somatic affixes.3

The feature of somatic prefixes that we wish to concentrate on in this section is that, although each somatic prefix has a clear prototypical meaning, as given in (11), nonetheless their usage is extended semantically, gradually but potentially well beyond the original meaning. Some of these extensions may be still synchronically valid, while others are perhaps only reflections of a diachronic process of extension. Table 1 sets out in diagrammatic form the extensions of the somatic prefix aka- ‘mouth’, while (12)–(33) provide illustration of lexical items with this prefix, in some cases including comparison with the same root occurring with a different somatic prefix or with no somatic prefix. Table 1 is to be read from left to right, without crossing lines, although the examples that follow exemplify explicitly each of the permitted extension paths.

Figure 3.Table 1. Extensions of somatic prefix aka- ‘mouth’.

Example (12) illustrates the literal use in the sense of ‘mouth’, including in the lexical item for ‘mouth’ itself, whose root also occurs without a somatic prefix meaning ‘to make a hole in the ground’. Also included are parts of the mouth and actions relating to the mouth.



aka-baŋ ‘mouth’

cf. baŋ ‘to make a hole in the ground’ (Z&C, p. 101)

aka-mewadi ‘to shut the mouth’

cf. mewadi ‘to shut’ (Z&C, p. 102)

aka-etal ‘tongue’ (Z&C, p. 101)

The ‘mouth’ prefix can be extended to adjacent parts of the body, such as chin, throat, and ear, as in (13)–(15).


mouth > chin

aka-ada ‘chin’ (Z&C, P. 101)

aka-talatim ‘hairless on the chin’ (Z&C, p. 102)


mouth > throat

aka-nilib ‘to tie tight round the throat’

cf. nilib ‘to tie tightly’ (Z&C, p. 102)


mouth > ear

aka-ya ‘inside of ear, gill’ (Z&C, p. 102)

Since the mouth is used for tasting and eating, it can be extended to these two semantic domains, as in (16)–(17), and also to food, as in (18).


mouth > taste

aka-beriŋa ‘tasty’

cf. beriŋa ‘good (of non-human referent)’ (Z&C, p. 103)


mouth > eat

aka-tau ‘to eat turtle’

cf. tau ‘hawksbill turtle’ (Z&C, p. 103)


mouth > food

aka-boka ‘morsel (of food)’ (Z&C, p. 103)

The concept food in turn leads to further extensions involving cooking/eating utensils and edible things/food sources, as in (19)–(20).


mouth > food > cooking/eating utensil

aka-nalama ‘clean (of cooking/eating utensil)’

cf. nalama ‘clean’

aka-uya ‘to heat (cooking pot); hot (of water)’

cf. uya ‘hot, warm (of weather)’ (Z&C, p. 104)


mouth > food > edible thing/food source

aka-ban ‘pea’

cf. ig-ban ‘seed’ (Z&C, p. 104)

Food source leads further to the concept tree, since trees were the source of much of the food regularly consumed by the hunter-gathering Akabea community, whence (21); and to ‘flower’, since flowers typically grow on trees in the Andamans.


mouth > food > edible thing/food source > tree

aka-taŋ ‘tree’

aka-kɔl ‘flower’ (Z&C, p. 104)

Tree provides a further path to wood, undergrowth, and canoe, the last given that the traditional Akabea canoe was a dugout tree, as in (23)–(24).


mouth > food > edible thing/food source > tree > wood

aka-tarali ‘to split in half (e.g. piece of wood)’

cf. tarali ‘to slit, split’ (Z&C, p. 105)


mouth > food > edible thing/food source > tree > undergrowth

aka-mal ‘to push aside undergrowth’

cf. ot-mal ‘to part (hair)’ (Z&C, p. 105)


mouth > food > edible thing/food source > tree > canoe

aka-kag ‘to approach shore by canoe’

cf. ar-kag ‘to visit by water’ (Z&C, p. 105)

Canoe in turn provides a link to watercraft in general, and to sailors, i.e. those who crew watercraft, as in (25)–(26). While there is no direct link between ‘mouth’ and ‘sailor’, the chain of semantic extensions shows that an apparently anomalous use of the ‘mouth’ prefix can nonetheless be motivated cognitively.


mouth > food > edible thing/food source > tree > canoe > watercraft

aka-birma ‘steamship’

cf. birma ‘gun; ship’s funnel’ (Z&C, p. 105)


mouth > food > edible thing/food source > tree > canoe > sailor

aka-ʤuru ‘sailor, fisherman’

cf. ʤuru ‘sea’ (Z&C, p. 105)

Backtracking to tree, this concept provides a basis for ‘arrow/spear/harpoon’, like a canoe made traditionally from a tree, as in (27)


mouth > food > edible thing/food source > tree > arrow/spear/harpoon

aka-tar-tel ‘to fit harpoon head to shaft’

cf. ar-tar-tel ‘to fit waist belt to body’ (Z&C, p. 106)

And backtracking all the way to mouth, we find items referring to the other main function of the mouth, speech, in (28), with further extension to noise in general in (29). Note that the ethno-linguistic term aka-bea ‘Akabea’ identifies in terms of speech, with an extension to ethnicity (i.e. the speakers of the language of the spring water); all of the names of the ten Great Andamanese ethno-linguistic groups use a cognate of this prefix (see (2)).


mouth > speech

aka-tegi ‘voice, language’

cf. tegi ‘sound’

aka-bea ‘Akabea’

cf. bea ‘spring water’

aka-mulua ‘deaf’

aka-dun ‘to groan’ (Z&C, p. 106)


mouth > speech > sound

aka-yeŋ ‘sound of surf’ (Z&C, p. 106)

Finally, there are extensions based on physical similarity to the mouth and its position on the face, namely ‘orifice’ in (30)–(31) and ‘front part’ in (32).


mouth > orifice

aka-kɔr ‘hole’

cf. kɔr ‘circle’ (Z&C, p. 107)


mouth > orifice > container

aka-kalaka ‘to open (a box)’

cf. ot-kalaka ‘bare’

aka-ela ‘to fill (vessel) with water’

cf. ela ‘to pour, bale out’ (Z&C, p. 107)


mouth > front part

aka-ʧeka ‘front part’ (Z&C, p. 107)

aka-papia ‘to wave hands in front of body’

cf. iʤ-i-papia ‘to flap (as bird’s wing)’ (Z&C, p. 108)

There are, however, a few residual instances where the semantic extension path is unclear, as in (33). These may reflect instances where our ignorance of the details of Akabea culture means that we cannot reconstruct a plausible link, or are confronted with a number of possible links of equal plausibility.


aka-keli ‘to go round a corner’

cf. ot-keli ‘to go round a small island’ (Z&C, p. 108)

4. Verb root ellipsis

The other phenomenon that we want to look at is one that we have called “verb root ellipsis”. We can start off with a quote from the literature on morphological theory.

“It is a common intuition that Ø-morphemes are only possible for very specific types of morphemes. They are expected to occur as affixes, but not as roots. Moreover, Ø seems more likely with functional material (whether affixal or not) than with lexical morphemes.” (TROMMER, 2012, p. 353)

So zero morphemes are only expected for very specific types of morphemes, for affixes as opposed to roots, for grammatical as opposed to lexical morphemes. We are going to be concerned with zero morphemes that are lexical roots.

There are at least two different phenomena that are potentially concerned here. The first one, which is not directly relevant here, is that some languages have lexical zero roots for a particular lexical item, i.e. zero contrasts with phonologically overt sequences for other roots. This is a rather rare phenomenon but certainly well attested cross-linguistically. An example from the Bardi language of Australia is given in (34). Here, the verb form consists of a first person prefix and a tense suffix, but no root – and this is the general form of the root ‘give’ in this language.

Bardi (Nyunyulan family):




‘I gave’ (BOWERN, 2012, p. 579)

What we are going to be concerned with in this section are instances of zero roots that arise through ellipsis. Now of course English also has a construction that we might think of as verb ellipsis. Under certain circumstances of identity across discourse, one can omit the second occurrence of the verb in (35) to give (36). However, note that what happens is ellipsis of the whole word singing, i.e. both the root sing and the suffix -ing. What English does not allow is just dropping the root and retaining the suffix, as in (37).

(35) Lee is singing, and Kim is singing too

(36) Lee is singing, and Kim is — too

(37) *Lee is singing, and Kim is —ing too

What we find in Akabea and some other Great Andamanese languages is precisely the construction as in (37).4 Several subtypes of this usage can be distinguished.

First, we sometimes find an ellipted verb root that is retrievable from the preceding conversational turn, as in the mini-dialogue in (38)–(39). One person says to eat some pork now, and the other person replies that they will only do so in the evening. In English one can omit eat, but this is omission of a whole word. What we find in the Akabea first turn in (38) is the verb ‘eat’ in the non-past, i.e. a lexical root with a tense suffix. The reply (38) has the non-past suffix directly following the pronoun expressing the subject, but the verb root is now zero – all that is left of the verb is the suffix, a suffix with no root, certainly no overt root.


aʧitek reg dama mek-ke

now pig flesh eat-nPST

‘Eat some pork now!’


yaba=da / wai dila=len d-o Ø-ke


‘No (lit. (it) is not). I will in the evening’ (Z&C, p. 152)

Second, the motion verb ɔn ‘come, go’ can be omitted if it is clear from the context, as in (40). Here the root would presumably have been possible, but all that we actually find is the non-past suffix, so again we have a suffix but no root. The previous word expresses the destination through the allative suffix attached to the nominalization of the verb ‘hunt’, so one can work out the meaning from the linguistic context, but there is no actual root there.


an ŋ-ar-at-duru dele-ŋa=lat Ø-ke


‘Will you all [go] hunting?’ (Z&C, p. 186)

One might compare this with the kind of ellipsis one sees in German with modal verbs, as illustrated in (41)–(42). The verb expressing motion can be ellipted in German when it is dependent on a modal verb, but crucially the whole word must be ellipted, it is not possible just to omit the root and leave the infinitive suffix dangling.


Kim muss nach Hause geh-en

Kim must.PRS.3SG to home go-INF

‘Kim must go home’


Kim muss nach Hause

Kim must.PRS.3SG to home

‘Kim must [go] home’

Third, the verb perek ‘strike, hit’ appears apparently context-freely as a zero root, although the number of examples attested is rather small. In (43)–(44) we have a minimal pair, where in (43) the verb root is present, while in (44) it is absent. The morphology of the verb is somewhat more complex, so these examples merit more detailed discussion. Both start off with an independent subject pronoun, series II because the tense is non-past. The second phonological word starts off with a series VII proclitic pronoun, here indexing the direct object, followed by the somatic prefix for ‘head’ indicating the locus of the action. Version (43) then follows with the root ‘strike’ and ends with the non-past tense suffix. Version (44) differs only in the ellipsis of the root, i.e. the second phonological word consists of a proclitic, a prefix, and a suffix, but no root. A more literal translation of (44) into English might be something like ‘I will you on the head’.


d-o ŋ=ot-perek-ke

1SG-II 2.VII=SP-strike-nPST

‘I will strike you on the head’


d-o ŋ=ot-Ø-ke


‘I will [strike] you on the head’ (Z&C, p. 153)

Verb root ellipsis is a typologically very rare phenomenon. If the only evidence were from the documentation of Akabea and other Great Andamanese languages, one might well object that there might be a problem with the documentation, perhaps these documenters who were not linguists just got it wrong. But as it happens we do have a couple of other clear attestations plus one that may be a borderline case. One clear parallel is Inuktitut (SWIFT & ALLEN, 2002), more specifically the variety spoken in northern Quebec. In this language we find the construction well attested in conversational data, with the possibility of omitting the verb base (i.e. the root plus in some cases some suffixes), leaving behind dangling suffixes. In Kwaza, a language isolate of Brazil, under appropriate conversational conditions the root of a verb can be omitted if its sense is retrievable (VAN DER VOORT, 2004), again leaving dangling affixes. A further possible example is the Australian language Jingulu (PENSALFINI, 2003), although here the data are not so clear and it may be that what is going on is rather omission of part of a discontinuous stem. For a fuller treatment of verb root ellipsis, both in Akabea and elsewhere, reference should be made to Comrie & Zamponi (2019b).

Figure 3 shows the geographical distribution of verb root ellipsis in the current state of our knowledge, with one dot standing for Akabea and other Great Andamanese languages with this phenomenon, and including the less clear case of Jingulu. Probably the only thing one can say about the geographic distribution is that it is not areal – there are four languages not quite as far apart from one another as they could conceivably be but coming close to that. And of course that the phenomenon is cross-linguistically rare.

Figure 4.Figure 3. Verb root ellipsis.

5. Epilogue

We have tried to tell our audience something about the Akabea language, but a reasonable response at this point might be to say: OK, that's fine, but this language was spoken by people, it was spoken by a community – what about that community? Well, the situation with respect to Akabea is different from that often encountered in many studies of endangered or recently extinct languages. In such situations one is accustomed to be dealing with language shift, with the survival of a community but not of its language. However, as already noted in the Introduction this is not the case with Akabea, in contrast to North Andamanese and its most recent manifestation, Present-day Great Andamanese. The demographic collapse of the Akabea led to the complete physical disappearance of the community. There is now no Akabea community, there are no Akabea individuals. But given that we have been using the material from their language, what can we say in response to our social responsibilities? First of all, we cannot claim to speak for the Akabea. The documentation we have was done exclusively by outsiders and it reflects the interests of those outsiders, how they wanted to portray the Akabea people, how they wanted to portray their language. At best what we can do is to try and ensure that the Akabea language takes its rightful place as part of humanity’s cultural heritage and the Akabea linguistic material finds its rightful place in linguistic science.

We write these words as the COVID-19 pandemic continues to rage. Comparison with the demographic collapse of the Akabea is very relevant. As already noted, the prime cause of this demographic collapse was a health issue. Introduced diseases to which the indigenous people had no immunity decimated and in some cases wiped out whole communities. Radcliffe-Brown wrote in The Andaman and Nicobar Islands: Local Gazetteer for 1908 that the root cause was “disease, introduced by the carelessness and callousness of individuals”. The colonial administration had failed to take precautions to prevent the introduction and spread of disease, and when faced with epidemics among the indigenous population sometimes provided health care, but often had other priorities, in particular the efficient running of the penal colony. There is a lesson we can draw in the midst of the current pandemic, namely that we should indeed learn from history and that history should not simply be a chronicle of the repetition of the same errors with ever more devastating consequences.


1, 2, 3 grammatical persons
I, II, III, IV, VII pronoun series
ALL allative
CAUS causative
COP copula
DEF definite
FOC focus
INF infinitive
LOC locative
LROOT ellipted root
NEG negative
NMLZ nominalizer
nPST non-past
PL plural
PRET preterit
REL relative
REMPST remote past
SG singular
SP somatic prefix
Y/N polarity question
Table 1.


  1. ABBI Anvita. Dictionary of the Great Andamanese language: English – Great Andamanese – Hindi. Ratna Sagar: Delhi; 2012.
  2. ABBI Anvita. A grammar of the Great Andamanese language: An ethnolinguistic study. Brill: Leiden; 2012.
  3. BARIK S. S., R. SAHANI B. V. PRASAD, P. ENDICOTT M. METSPALU, B. N. SARKAR S. Bhattacharya, P. C. Annapoorna J. Sreenath, D. Sun J. J. Sanchez, S. Y. Ho A. Chandrasekar, V. R. Rao. Detailed mtDNA genotypes permit a reassessment of the settlement and population structure of the Andaman Islands. American Journal of Physical Anthropology. 2008; 136:19-27.
  4. BOWERN Claire Louise. A grammar of Bardi. De Gruyter Mouton: Berlin; 2012.
  5. COMRIE Bernard, ZAMPONI Raoul. Subgrouping and lexical distance in the Great Andamanese family. In: KOTIN Michail L.. Wortschätze und Sprachwelten: Beiträge zu Sprachtypologie, kontrastiver Wort- bzw. Wortschatzforschung und Pragmatik. Lang: Lang; 2019a:35-57.
  6. COMRIE Bernard , ZAMPONI Raoul. Verb Root Ellipsis. In: BAERMAN Matthew, BOND Oliver, HIPPISLEY Andrew. Morphological perspectives: Papers in honour of Greville G. Corbett. Edinburgh University Press: Edinburgh University Press; 2019b:233-280.
  7. Haspelmath Martin, Dryer Matthew S., COMRIE Bernard, GIL David. The world atlas of language structures. Oxford University Press: Oxford; 2005.
  8. MAN Edward H.. Dictionary of the South Andaman language. Indian Antiquary. 48:Supp., 1919, pp. 1-84; v. 49, Supp., 1920, pp. 85-136, 50, Supp., 1921, pp. 137-164, v. 51, Supp., 1922, pp. 165-188, v. 52, Supp., 1923, pp. 189-203..
  9. PENSALFINI Rob. A grammar of Jingulu, an Aboriginal language of the Northern Territory. Pacific Linguistics: Canberra; 2003.
  10. PORTMAN Maurice V.. Manual of the Andamanese languages. W. H. Allen: London; 1887.
  11. PORTMAN Maurice V.. Notes on the languages of the South Andaman group of tribes. Office of the Superintendent of Government Printing: Calcutta; 1898.
  12. SWIFT Mary D., ALLEN Shanley E. M.. Verb base ellipsis in Inuktitut conversational discourse. International Journal of American Linguistics. 2002; 68:133-156.
  13. TROMMER Jochen. Ø-exponence. In: TROMMER Jochen. The morphology and phonology of exponence. Oxford University Press: Oxford University Press; 2012:326-354.
  14. VAN DER VOORT Hein. A grammar of Kwaza. Mouton de Gruyter: Berlin; 2004.
  15. ZAMPONI Raoul, COMRIE Bernard. A grammar of Akabea. Oxford University Press: Oxford; 2020.