Aspects of Garifuna attributive possession in

comparative perspective

ALT 10, Leipzig 2013

Steffen Haurholm-Larsen
Institut für Sprachwissenschaft
Universität Bern
[email protected]
August 17, 2013

Classification Arawak / Maipurean

Location Caribbean seaboard of Northern Central America
Data sources Author's fieldwork in Honduras (ongoing), otherwise cited
Special characters <ü> = [ɨ], <ny> = [ɲ], <r> = [ɾ] / [ʁ], <ch> = [ʧ], <y> = [j]
Stress Stress is word-initial unless otherwise marked

North Arawak speaking area (Map adapted from Kaufman (2007) through www.llmap.org)

I will not go into detail regarding the history of Garifuna but simply say that Gari-
funa has been heavily influenced by Carib, more so than the contact influence in other

1
North Arawak languages, due to a pre-colonial situation of bilingualism that dramati-
cally changed the lexicon and parts of the grammar. What emerged from this situation
was an Arawak language but with dedicated men's speech associated with Carib lexicon
and women's speech associated with Arawak lexicon. This gendered speech practice,
however, is today all but gone.

1 Introduction
Among the languages that have noun classifiers, only some have classifiers in pos-
sessive constructions. In those languages that do have possessive classifiers, charac-
teristics about three different aspects of the possessive relation may be classified: 1)
possessor, 2) possessed and 3) the relationship that holds between possessor and pos-
sessed (Aikhenvald, 2000, p. 17). In this talk I will focus on the last of the three which
I label relational classifiers following Lichtenberk (1983)1 . I will take as point of de-
parture my own data from ongoing fieldwork on Garifuna and the Garifuna data will
be compared to Arawak, Carib, and Oceanic languages that also have relational clas-
sifiers. The aim of this comparison will be to look for evidence that the classifiers of
the South American languages are less grammaticalized than the Oceanic ones, as has
been claimed by Grinevald (2000, p. 81) in (1).
(1) South American systems ... [of relational classifiers] ... are in fact said to
be emergent systems, less grammaticalized than the Micronesian systems, with
much discourse sensitivity in the use of the classifier construction.
However, I will start by giving a small introduction to Garifuna attributive posses-
sion.

2 Attributive possession and relational classifiers in Garifuna

2.1 Attributive possession in Garifuna

Garifuna nouns are possessed by adding a prefix to the possessed noun that indexes
person and number of possessor. The possessive paradigm is shown in (2) and examples
are given in (3)2 .

SG PL
1 n- w(a)-
(2) 2 b- h-
3.M l- h(a)-
3.F t- h(a)-

1 Relational classifiers have also been called "Genitive" and "Possessive" but these terms only say some-

thing about the syntactic context in which they appear while failing to express their function.
2 Abbreviations: F = feminine, M = masculine, CL = classifier, ABS = absolute, POSS = possessive

2
(3) a. n-uguchi
1.SG-father
'my father'

b. t-i'chügü
3.SG.F-head
'her head'

There is a distinction between inalienably and alienably possessed nouns. Inalien-

able covers kinship terms, body parts and other part-whole relations while alienable
covers all other nouns. Alienably possessed nouns are differentiated from inalienable
ones by the use of a possessive suffix - examples are shown in (4).
(4) a. nu-'baruru-n
1.SG-platain-POSS
'my plantain'

b. n-eyfi-te
1.SG-bean-POSS
'my beans'

c. ni-duna-ri
1.SG-water-POSS
'my water'

It is unclear what semantic content or grammatical function the difference between pos-
sessive suffix has (or used to have).

A null suffix may be posited as possessive suffix for inalienably possessed nouns;
this is manifested in a vowel change which happens regularly at suffixation - this is
illustrated in (5) and (6).
(5) a. u'bareü
'finger nail'

b. n-ubara-Ø
1.SG-nail-POSS
'my finger nail'

c. u'ri:reü
'breast'

d. tu'-ri:ra-Ø
3.SG.F-breast-POSS
'her breast'

3
(6) a. u'madeü
'friend'

b. u'mada-gu
friend-PL
'friends'

Number of possessed is marked by a suffix as shown in (7). Number marking in Gari-

funa is only a feature of nouns with animate referents.
(7) a. n-i'bari-nu
1.SG-grandchild-PL
'my grandchildren'

b. n-ibiri-gu
1.SG-brother-PL
'my brothers'

c. n-i'biri-nya
1.SG-relative-PL
'my relatives'

d. l-i'rahü-nyü
3.SG.M-child-PL
'his children'

In most Arawak languages, inalienably possessed nouns have absolute (unpossessed)

forms which are marked by a suffix *-ʧi or *-hVi (Aikhenvald, 1999, p. 82).
Garifuna has a handful of different absolute -CV suffixes, but they have a very
restricted distribution in the lexicon as most nouns lack them; examples showing the
attested forms the absolute suffix are given in (8)

(8) a. uguchu-ru
mother-ABS
'mother'

b. n-uguchu
1.SG-mother
'my mother'

c. uguchi-li
father-ABS
'father'

4
 d. n-uguchi
1.SG-father
'my father'

e. u'gufera-ni
godfather-ABS
'godfather'

f. n-u'gufera
1.SG-godfather
'my godfather'

g. a'naga-ni
back-ABS
'back'

h. n-a'naga
1.SG-back
'my back'

i. e-un
penis-ABS
'penis'

j. n-e
1.SG-penis
'my penis'

Considering the restricted distribution of absolute suffixes, it is possible that these

are vestiges of an earlier situation before the emergence of possessive suffixes, which
now carry the workload of distinguishing alienability. In other words, if Garifuna had
no possessive suffix, there would be no formal marking of alienability and there would
be a need for alternative means of marking this distinction.

2.1.1 Unpossessable nouns

A few nouns are unpossessable, that is, they cannot take possessive morphology - in-
stead, they have suppletive equivalent lexemes used to express possession as shown in
(9) and (10).
(9) a. muna
house
'house'

5
 b. *nu-muna-n
1.SG-house-POSS
1sg-house
'my house'

c. n-uba-n
1.SG-house-POSS
'my house'

(10) a. e'reba
cassava.bread
'cassava bread'

b. *n-e'reba-te
1.SG-cassava.bread-POSS
'my cassava bread'

c. n-ubu-te
1.SG-cassava.bread-POSS
'my cassava bread'

Another small portion of nouns is also non-possessable. However, they take possessed
forms of other nouns which themselves have an unpossessed form, as shown in (11)
and (12).
(11) a. u'bow
'village'

b. *n-u'bow
1.SG-village
'my village'

c. gu'ruyara
'canoe'

d. *nu-'guruyara
1.SG-canoe
'canoe'

(12) a. a'geyra
'village'

b. n-a'geyra
1.SG-village
'my village'

6
 c. u'guney
'canoe'

d. b-u'gune
2.SG-canoe
'your canoe'

Of course an interesting question would be whether or not these are semantically true
equivalents - I have yet to answer this question. I note, however, that (11) and (12) are
inalienably possessed as they have no suffix.

These unpossessable nouns might be relevant to the discussion that follows here,
because in many languages, relational classifiers are used as a means of possessing
unpossessable nouns - however, the nouns in (11) and (12) do not take classifiers in
Garifuna and I have yet to figure out how they might fit into the overall scheme of
things.

2.2 Relational classifiers in Garifuna

Lexical items of certain semantic categories are possessed, not by direct affixation as
we have seen above, but by the use of a separate word preceding the possessed noun,
namely a relational classifier which classifies the relation that holds between Possessor
and the Possessed.

Relational classifiers are different from all other classifiers which typically classify
inherent attributes of nouns (shape, texture, animacy) although some relational classi-
fiers also encode information about the Possessed (cf. the Food (non-meat) vs. Food
(meat) distinction in (13) and (14).)

The relational classifier takes the possessive morphology in stead of the possessed
noun.

As shown in the examples that follow, Garifuna has 5 different relational classifiers:
FOOD (non-meat), FOOD (meat), DRINK, DOMESTIC ANIMAL / PET, GENERAL.

7
(13) -eygan Food, (non-meat)
a. n-eygan faluma
1.SG-CL coconut
'my coconut'

b. n-eygan hu'dutu
1.SG-CL banana.mash
'my banana mash'

c. b-eygan da'rasa
2.SG-CL tamale
'your banana tamale'

(14) -uyi Food, (meat)

a. n-uyi udereü
1.SG-CL fish
'my fish'

b. n-uyi gwa'yamaga
1.SG-CL iguana
'my iguana'

(15) -niye Drink

a. nu-'niye faluma
1.SG-CL coconut
'my coconut'

b. bu-'niye gu'retu
2.SG-CL gruel
'your banana-manioc gruel'

(16) -lügün Domestic animal / Pet

a. bi-'lügün ga'bayu
2.SG-CL horse
'your horse'

b. li-'lügün bagasu
3.SG.M-CL cow
'his cow'

(17) -ani General possession

a. n-ani faluma
1.SG-CL coconut
'my coconut'

8
 b. n-ani bweybwey
1.SG-CL wild.grape
'my wild grape'

c. n-ani-gu a'feyndihati-nyu
1.SG-CL-PL painter-PL
'my painters'

The noun faluma 'coconut' serves as a good illustration of the discourse dependent
nature of relational classifiers, as it appears as both Food (non-meat) (13), Drink (15)
and General possession (17); so which classifier is used really depends on the intent of
possessor, albeit within the pragmatic boundaries of the possessed - a stone, e.g. can
hardly be destined for drinking.

The most uncertainty that I have experienced is with the DOMESTIC ANIMAL
classifier - normally if you ask if one can say ni-'lügün fudi 'my cockroach' you would
be told that this is not possible because nobody will have a cockroach as a pet - but
then I have a narration involving a man in prison who has a cockroach as a pet and then
suddenly in this context (18) is possible.

(18) a. li-lügün fudi

3.SG.M-CL.PET cockroach
'his pet cockroach'
Some relational classifiers have alternative classifier forms for the first person singular,
as shown in the following examples.

(19) a. n-uyi gwa'yamaga

1.SG-CL iguana
'my iguana'

b. yudi gwa'yamaga
1sg.CL.FOOD iguana
'my iguana'

(20) a. nu-'niye a'dulu

1.SG-CL.DRINK gruel
'my porridge'

b. yagu a'dulu
1sg.CL.DRINK gruel
'my porridge'

9
(21) a. ni-'lügün mesu
1.SG-CL.PET cat
'my cat'

b. yegü mesu
1sg.CL.PET cat
'my cat'

Formally, these classifiers seem to haven little if anything in common with the reg-
ular relational classifiers presented above - the only regularity found here is that they
are all CVCV with an initial y- and as far as I know, there is no such prefix in Garifuna.

There is, however, a 1.SG possessive y- prefix in Carib languages and it would seem
that Garifuna has borrowed the classifiers in the complex shape 1.SG y- + CLASSIFIER
as monomorphemic words, cf. the Panare examples in (22).

(22) Panare (Payne and Payne, 2012)

a. y-ëwa-n
1.SG-nose-POSS
'my nose'

b. yu-pïj
1.SG-wife
'my wife'

According to speakers, the choice between a regular 1.SG classifier and the alternate
form is arbitrary, but it might be that the difference used to be one of men's speech vs.
women's speech.

Finally, it is grammatically possible to possess e.g. foods or animals without the use
of a classifier as shown in (23), but there seems to be a preference for using classsifiers.

(23) a. nu-'bagasu-n
1.SG-cow-POSS
'my cow'

b. nu-'baruru-n
1.SG-banana-POSS
'my banana'

c. ni-'so:pa-n
1.SG-soup-POSS
'my soup'

10
I have yet to dig deeper into the semantic or pragmatic constraints that ultimately decide
between the use between the GENERAL classifier -ani and direct possessive morphol-
ogy.

3 Relational classifiers comparatively

3.1 Arawak
After having seen the relatively elaborate system of relational classifiers in Garifuna,
it may come as a surprise that in the Arawak language family this phenomenon is ex-
tremely restricted; relational classifiers are only attested in a few of the North Arawak
languages: Palikur, Bahwana and Achagua (Aikhenvald, 1999, p. 84) (and possibly
Baniwa of Içana) - examples are shown below.

All of the Arawak languages discussed here mark inalienable possession, including
kinship terms, bodyparts and part-whole relations, by direct possessive morphology,
but they differ as to whether or not alienably possessed nouns can take such direct pos-
sessive marking.

Achagua has two relational classifiers: one for DOMESTIC ANIMAL and one for
GENERAL possession - as the examples in (24) show, like Garifuna, Achagua has the
possibility of using either direct possessive morphology or a classifier. According to
Meléndez, using direct possessive morphology might express a closer relationship than
when using a classifier.
(24) Achagua (Meléndez, 1998)
a. nu-ʃina: ku'lupa
1.SG-CL.GENERAL fish.hook
'my fish hook'

b. nu-'kulupa-ni
1.SG-fish.hook-POSS
'my fish hook'

Note that the GENERAL classifier in Achagua does not seem to have anything formally
in common with any of the Garifuna classifiers - in fact, none of the North Arawak re-
lational classifiers that I have seen are cognate.

Palikur has 5 relational classifiers: FOOD (non-meat), FOOD (meat), PET, PLANT,
CHILD - two of them are shown in (25).

11
(25) Palikur (Aikhenvald and Green, 1998)
a. nu-mutra pilatno
1.SG-CL.PLANT banana
'my (planted) banana'

b. nu-mana pilatno
1.SG-CL.FOOD banana
'my banana (for eating)'

In Palikur, those nouns which take a classifier are unable to take direct possessive mor-
phology (Valadares, 2006); this is different from Garifuna and Achagua.

Again, the examples above show that the choice of classifier is determined by dis-
course context.

Bahwana only has 2 classifiers, one for DOMESTIC ANIMAL / PET and one for
GAME - it remains to be seen if nouns that take classifiers can also take possessive
morphology.

(26) Bahwana (Ramirez, 1992) cited in (Aikhenvald, 1999)

a. nu-üra haburu
1.SG-CL.PET parrot
'my parrot'

If indeed Bahwana turns out to only have 2 classifiers, the fact that they are both spe-
cialized is atypical cross-linguistically. In a language with only 2 relational classifiers,
we would expect one of them to be GENERAL: in the American 2-classifier systems
the other one tends to be DOMESTIC ANIMAL (Carlson and Payne, 1989) while in
Oceanic it tends to be FOOD (Aikhenvald, 2012, p. 20).

Considering the fact that relational classifiers are only found in a few North Arawak
languages, right in an area where the Arawak and Carib language families overlap, we
may entertain the hypothesis suggested by Aikhenvald (2000, p. 383) and Aikhenvald
and Green (1998, p. 460) that the North Arawak languages acquired relational classi-
fiers under influence of the Carib speakers in the area.

12
3.2 Carib
Inalienable possession in Carib works like in Arawak except that in Carib, possessor is
only indexed on the possessed noun if the latter is not expressed by a full NP - this is
examplified with data from Panare in (27).
(27) Panare (Carlson and Payne, 1989)
a. e'ñapa piya-n
Panare chief-POSS
'the chief of the Panare people'

b. a-piya-n
2.SG-chief-POSS
'your chief'

At least two Carib languages have relational classifiers, Macuxi and Panare.

In Macuxi, possession with relational classifiers is the only possessive strategy for
animals (Carlson and Payne, 1989). This is exempified in (28).

(28) Macuxi (Carlson and Payne, 1989)

a. iwarika u-y-eküng
monkey 1.SG-POSS-CL.PET
'my (pet) monkey'

b. baka itekung urə

cow CL.PET 1.SG.POSS
'my cow'

In addition, Macuxi has a FOOD classifier as shown in (29).

(29) séni azékari káno
this 2.SG.CL.FOOD guava
'this guava is for you'

It is unclear, however, whether other alienably possessed nouns take classifiers or not
- Carlson and Payne (1989) think that Macuxi might have a general classifier, and this
would then be used for all nouns not covered by the FOOD or DOMESTIC ANIMAL
classifiers.

Panare has the largest inventory of relational classifiers among the Carib languages;
they amount to more than 20. Alienable possessed nouns normally take a classifier -
examples are shown in (30).

13
(30) Panare (Matteí Muller, 1974); (Carlson and Payne, 1989)
a. y-uku-ng wanü
1.SG-CL.DRINK-POSS honey
'my honey (for drinking)'

b. y-empa-ng maranka
1.SG-CL.FRUIT-POSS orange
'my orange'

c. Ø-tëʔma-ng mwe
1.SG-CL.SOFT.FOOD-POSS egg
'my egg'

However, if an object is owned with no particular purpose in mind, according to Mat-

teí Muller (1974) it may be possessed by direct possessive morphology as if it were
inalienably possessed. But it seems to me that it is a difference in suffix that really
distinguishes alienability, just like we saw in Garifuna, cf. (31).
(31) Panare (Matteí Muller, 1974, p. 7-8) (Payne and Payne, 2012)
a. y-u-ng aʔmaka
1.SG-CL.FOOD-POSS manioc
'my manioc (for eating)'

b. y-aʔmaka-e
1.SG-manioc-POSS
'my manioc (still in the garden)'

Comparing (27-b) and (31-b) we see the suffixes -ng and -e, the former being used for
body parts and kinship terms, the latter for other possessed items.

This is consistent with other Carib languages that generally use possessive suffixes
to express alienability (Derbyshire, 1999, p. 41).

14
3.3 Oceanic
Like the American languages we have seen so far, Oceanic languages have two poses-
sive strategies: direct possessive morphology as illustrated by Kwaio in (32) is re-
stricted to inalienably possessed nouns including kinship terms, body parts and part-
whole relations.
(32) Kwaio (Lichtenberk, 1983)
a. oga-na boo
belly-3.SG pig
'a pig's belly'

b. asi-mu
younger.brother-2.SG
'your younger brother'

Alienble possessesion includes all other nouns and is indicated by the use of a rela-
tional classifier - the most common system is confined to 2 classifiers, one for FOOD
including both solid and fluid foods and another for GENERAL possession - these are
illustrated in (33) from Manam.
(33) Manam (Lichtenberk, 1983)
a. aine niu ʔana-gu
woman coconut CL.FOOD-3.PL
'the women's coconuts'

b. ʔusi ne-gu
loincloth CL.GENERAL-1.SG
'my loincloth'

As was the case in the American languages, the choice of classifier is entirely dependent
on pragmatics, i.e. the relation that holds between possessor and the possessed or the
intent of possessor regarding possessed. This becomes evident by the fact that different
classifiers can be used for the same lexeme as shown in (34) and (35) from Fijian and
Marshallese.
(34) Fijian (Lichtenberk, 1983)
a. na me-qu yaqona
art CL.DRINK-1.SG kava
'my kava (for drinking)'

b. na no-qu yaqona
art CL.GENERAL-1.SG kava
'my kava (for selling)'

15
(35) Marshallese (Lichtenberk, 1983)
a. kötka-m mä
CL.PLANT-2.SG breadfruit
'you breadfruit tree (which you planted)'

b. neji-m mä
CL.VALUABLE-2.SG breadfruit
'your breadfruit tree (not planted by you)'

c. kijö-m mä
CL.FOOD-2.SG breadfruit
'your breadfruit (for eating)'

One major difference between the Arawak and Carib classifier systems on the one
hand and the Oceanic ones on the other, is that in Oceanic all alienably possessed nouns
must take a classifier (Carlson and Payne, 1989, p. 90); (Lichtenberk, 1983, p. 150-151),
whereas in the American languages the picture is varied. In some Arawak and Carib
languages alienably possessed nouns have the additional possibility of direct affixation
- this alternate option may express either lack of specific purpose of the possessed as in
the Carib language Panare (Matteí Muller, 1974, p. 7-8) or a closer relationship between
possessor and possessed as we have seen in the Arawak langauge Achagua (Meléndez,
1998, p. 56-57).
The most straight forward explanation for this difference is that in directly pos-
sessed nouns, the alienability distinction is already marked by suffixation, making the
presence or absence of a classifier irrelevant with respect to alienability.

So, the obligatoryness of classifiers in some languages is not a sign of higher gram-
maticalization; because the choice of classifier is equally discourse dependent, the dif-
ference being the added workload of classifiers to distinguish between alienably and
inalienbly possessed nouns.

4 Conclusion
We have seen that languages with relational classifiers can be divided into two groups,
1) ones where all alienably possessed nouns MUST use a classifier when possessed, and
2) ones where the use of relational classifiers with alienably possessed nouns is only
one option which adds information about the relation between possessor and possessee;
in these languages, alienabilty is expressed by alternative means.

Interestingly, this is not a division between languages of the Americas on the one
hand and Oceanic languages on the other.

Moreover, American as well as Oceanic languages vary greatly in the size of clas-
sifier inventory but the normal situation is to have a small 2-classifier system.

16
References
Aikhenvald, Alexandra Y. 1999. The Arawak language family. In The Amazonian
Languages, chapter 3, 65--106. Cambridge University Press.

Aikhenvald, Alexandra Y. 2000. Classifiers: A Typology of Noun Categorization De-

vices. Oxford University Press.

Aikhenvald, Alexandra Y. 2012. Possession and Ownership: a cross-linguistic perspec-

tive. In Possession and Ownership - A Cross-Linguistic Typology, ed. Alexandra Y.
Aikhenvald and R. M. W. Dixon, 1--64. Oxford University Press, USA.
Aikhenvald, Alexandra Y., and Diana Green. 1998. Palikur and the Typology of Clas-
sifiers. Anthropological Linguistics 40:429--480.
Carlson, R., and Doris Payne. 1989. Genitive classifiers. In Proceedings of the Fourth
Annual Pacific Linguistics Conference, ed. Scott DeLancey Robert Carlsson, Spike
Gildea, Doris Payne, and Anju Saxena. Eugene: University of Oregon, 89-119.

Derbyshire, Desmond C. 1999. Carib. In The Amazonian languages, ed. Robert M. W

Dixon and A. Aikhenvald. Cambridge University Press.

Grinevald, Colette. 2000. A Morphosyntactic Typology of Classifiers. In Systems of

nominal classification, ed. Gunter Senft, number 4 in Language, culture, and cogni-
tion, 50--92. Cambridge ; New York: Cambridge University Press.
Kaufman, Terrence. 2007. South America. In Atlas of the World's Languages, ed. R. E.
Asher and Christopher Moseley, 83--93. Routledge, 2 edition.
Lichtenberk, Frantisek. 1983. Relational classifers. Lingua 60:147--176.

Matteí Muller, Marie-Claude. 1974. El sistema de posesion en la lengua panare.

Antropologica 38:3--14.
Meléndez, Miguel Ángel. 1998. La lengua achagua: estudio gramatical. Bogotá:
Colciencias : Universidad de las Andes.
Payne, Thomas E., and Doris L. Payne. 2012. A Typological Grammar of Panare: A
Cariban Language of Venezuela. BRILL.
Ramirez, Henri. 1992. Le bahuana: une nouvelle langue de la famille arawak. Asso-
ciation d'Ethnolinguistique Amérindienne.
Valadares, Simoni M. B. 2006. Some Notes on Possessive Constructions in Palikur. In
Proceedings from the Ninth Workshop on American Indigenous Languages, ed. Lea
Harper, volume 18. University of California, Santa Barbara.

17