The Proto-Cariban and Tupí-Guaraní Object Nominalizing Prefix

Spike Gildea

University of Oregon
and
Museu Paraense Emilio Goeldi

Different higher-level classifications of the language families of South

America differ on the closeness of a relationship between the Cariban and
Tupí-Guaraní language families: Greenberg (1987) places Cariban and Tupían
languages in completely separate branches of Amerind; Loukotka (1967) links
Cariban and Tupían at a somewhat closer -- but still quite distant -- level;
Rodrigues (1985) argues for a close Cariban-Tupí relationship on the basis of
several apparent cognates between individual Cariban and Tupían languages.
In this paper I examine parallel syntactic patterns and a possible cognate
morpheme from the Cariban and Tupí-Guaraní families. On the basis of this
evidence, I tentatively support Rodrigues’ position, that Cariban and Tupían
are closely related.

The remainder of this paper is organized as follows: in §1 I discuss the

dominant nominalization strategy in both the Cariban and Tupí-Guaraní
families; in §2 I identify an exception to the dominant strategy in the Tupí-
Guaraní family; in §3 I show that a parallel exception and a possible cognate
morpheme exist in the Cariban family; and in §4 I discuss the implications of
these patterns for higher-level classifications.

1 The Dominant Nominalization Strategy in Cariban and Tupí-Guaraní

Jensen (1990:126-8) for Tupí-Guaraní and Gildea (1992:121-45) for Cariban
document virtually identical systems of nominalization: the verb bears a
nominalizing suffix and is possessed by its notional O or S (i.e. the
absolutive); the notional A (i.e. the ergative) cannot possess a
nominalization, but must occur in an oblique case if it occurs at all.1 This
strategy is illustrated in the following examples from Tupinambá (Jensen
1990) and Hixkaryana (Derbyshire 1985):

(1) Tupinambá (Tupí-Guaraní)

a. Action Nominalization Possessed by Notional S

né só-Ø
2Sg go-Nmlzr
‘your going’ (J’s 35, p. 126)

b. Circumstance Nominalization Possessed by Notional O

i-moján-áb
3-make-Nmlzr
‘its making place’ (J’s 38, p. 127)

1
(2) Hixkaryana (Cariban)

a. Action Nominalization Possessed by Notional S

teryeweryero ro-wanota-nɨ-r me kewehyaha

loudly 1-sing-Nmlzr-Pssn Denominalizer I-take-a-bath
‘I take a bath singing loudly’ (D’s 45b, p. 38)

b. Action Nominalization Possessed by Notional O

honkyo wo-nɨ-r xe wehxaha, Waraka wya

peccary shoot-Nmlzr-Pssn Desirous-of I-am Waraka by
‘I want Waraka to shoot Peccary.’ (D’s 45g. p. 39)

A further similarity between the families is that the absolutive possessor

can take the form of a free (pro)noun immediately preceding the possessed
nominalized verb (cf. 1a, 2b) or it can occur as a reduced pronominal clitic,
bound to the front of the verb (1b, 2a). However, it is important to note at
this point that despite the identity in systems, no cognates have been found
in either the nominalizing suffixes or the forms of the possessive
proclitics.

Although the ergative organization is relatively dominant, in each family

there is one idiosyncratic O nominalizer (i.e. an affix which, when attached
to the verb, derives a noun which refers to the notional O of the nominalized
verb). It differs from the other nominalizers in that (1) it is a prefix,
(2) it can co-occur with various other nominalizers, and (3) the resulting
derived nominal is obligatorily possessed by the notional A of the verb
rather than by the notional O. I examine this unique nominalizing prefix for
Tupí-Guaraní in §2 and for Cariban in §3.

2 The Proto-Tupí-Guaraní Object Nominalizing Prefix *emi-

Jensen (1990:127) describes the Proto-Tupí-Guaraní object nominalizing prefix
*emi- as idiosyncratic because “in all other cases, whether verbal
constructions or nominalizations, it is a suffix -- not a prefix -- which
creates the particular construction. Furthermore, there are no other cases
where Set 2 person markers [the possessive proclitics -- S.G.] reference A.”
Jensen offers three examples from Tupinambá:

(3) a. syé r-emi-'ú

1Sg Linker-O.Nmlzr-eat
‘that which I eat; my food’ (J’s 40, p. 127)

b. s-emi-moján-(a)
3-O.Nmlzr-make-(Nominal.Case)
‘that which he makes; his handiwork’ (J’s 41, p. 127)

c. syé r-emi-nupã
1Sg Linker-O.Nmlzr-hit
‘the one whom I hit’ (J’s 44, p. 128)

3 The Proto-Cariban Object Nominalizing Prefix *nï-

Gildea (1992:135) reconstructs the Proto-Cariban O nominalizing prefix as
*nɨ-, pointing out exactly the same idiosyncracies as Jensen notes for the
Proto-Tupí-Guaraní form.2 I now illustrate the cognate set for Proto-Cariban
*nï ‘O Nominalizer’ in 13 Cariban languages. For some languages, I offer
more discussion simply because more material is available to me. The nï-

2
prefix is a part of the system of nominalization in all languages for which
it has been described, hence it must be reconstructed as a part of the Proto-
Cariban nominal system.

Koehn and Koehn (1986:91) describe a complex object nominalizer in Apalaí:

“-ny-...-ry [is] Object resulting from an action.3 This occurs with
transitive stems only. The -ny- is preceded by possessor person markers.”
The ny- can also occur with the past tense nominalizer -hpyry. In 4a, the
stem mero 'write' is sandwiched between a ny- ‘O.Nmlzr’ prefix and a -ry
‘Possessed/Nmlzr’ suffix. The notional A is expressed as the first person
possessive prefix y- preceding the derived nominal. In 4b, the same verb
with the nominalizing prefix takes the past tense possessive/nominalizing
suffix and is possessed by the second person notional A.

(4) a. y-ny-mero-ry
1-O.Nmlzr-write-Nmlzr
‘the thing I am writing’ (K & K's 309, p. 91)
(lit. ‘my writee [written thing]’)

b. o-ny-mero-hpyry
2-O.Nmlzr-write-Nmlzr+Past
‘the thing you have written’ (K & K's 310, p. 92)
(lit. ‘your [past] writee [written thing]’)

Hoff (1968:137-8) describes a nasal prefix in Carib of Surinam which he

labels the ‘ni- category’. It is found only with transitive verbs, deriving
a nominal which is possessed not by the usual “who or what undergoes the
action”, but rather by “who or what performs it”. Hoff further states that
the derived nominal “can refer both to the action and to the person or thing
undergoing it.” Although in Carib, the derived nominal bearing the ni-
prefix does not always refer to the notional O, since that is one of the
synchronic functions, I continue to gloss it as such in the examples. The
ni- prefix in Carib can co-occur with the following nominalizing suffixes
(from the charts and discussion, Hoff 1968:148-54): -Ø, -rï ‘Possession’,
-xpo ‘Perfective’, -mbo ‘former’, -xto ‘Negative Desiderative’, -xpo-to
‘Perfective-when’.

The following examples are taken from texts in Hoff (1968). In 5, the verb
i'mo ‘break’ bears the O Nominalizing prefix n- and the perfective
nominalizer -xpo, then is possessed by the first person notional A Ø-. In 6,
the verb aaro ‘take’ bears the same n- prefix and -xpo suffixes, but is
possessed by the second person notional A, a- ‘2’. In 7 the verb paato
‘cross’ bears the ni- prefix, the -rï ‘Possessive’ suffix, and the i- ‘3’
personal possessive prefix. In addition, the derived nominal bears the
plural suffix -koN, referring to the possessor (cf. Hoff 1968:248 for
discussion of combining possession and plural morphology). In all cases, the
derived nominal functions as a restrictive object relative clause.

(5) Ø-ni'móópo ro xkuru mooro

Ø-n-i'mo-xpo ro xkuru mooro
1-O.Nmlzr-break-Perfective.Nmlzr indeed beforehand that

ataamur(u) enuuru
a-taamuusï-rï Ø-oonu-rï
2-grandfather-Possessive Ø-eye-Possessive
‘your grandfather's eye that was broken by me?’ (H, p. 324-5)
(lit. ‘my breakee before, that your grandfather’s eye?’)

3
(6) moosé anaaróóp(o) oomïyááko pooko
moose a-n-aaro-xpo oomïyaako pooko
this 2-O.Nmlzr-take-Perfective.Nmlzr young.woman with
‘with this girl who was carried off by you’ (H, p. 322-3)
(lit. ‘with this your takee girl’)

(7) iniipáátorîîkoN
i-n-paato-rï-:koN
3-O.Nmlzr-cross-Possessive-Plural
'the one that they crossed.' (H, p. 328-9)
(lit. ‘their crossed one’)

For Hixkaryana, Derbyshire (1985) devotes only one sentence to a construction

-nɨ- ... -nɨ- ‘object resulting from action’. In this construction he states
that “the direct object... is the central part of the derivation, and the
possessor prefixes in this (unique transitive) case mark the subject of the
action.” In an appendix (pp. 232-3) the following nominalizing suffixes are
listed as co-occurring with the -nɨ- ‘object resulting from action’
prefix:-nɨ-rɨ ‘Present’, -hɨto-rɨ ‘Negative’, -thɨ-rɨ ‘Past’, -hɨto-rɨ tho
‘Negative Past’. In 8, the verb nyake ‘send’ bears the nɨ- prefix, the
action nominalizing suffix -nɨ, the possessive suffix -rɨ; it is possessed by the
preceding noun Waraka.

(8) Manawasɨ hona Waraka Ø-nɨ-nyake-nɨ-rɨ uhutwehe

Manaus to Waraka 3-O.Nmlzr-send-Action.Nmlzr-Possession I-know-him
‘I know the one whom Waraka sent to Manaus.’ (D’s 65b, p. 49)
(lit. ‘I know Waraka's sendee to Manaus.’)

In Kuikúro, Franchetto (1990) shows convincing evidence that a verb

bearing the *nɨ- prefix is primarily used on as a main clause verbal
inflection. She is speaking of this main clause verbal inflection as
she describes the now-familiar morphosyntax associated with the *nɨ-
prefix (p. 412):

...de-ergativization applies to inherently transitive verbs and

results in the verb agreeing with the actor or agent of the event
described by the verb. De-ergativization is formally marked by
the verbal prefix ñ- or g- [N]; the prefixed verb occurs
immediately after the unmarked subject. The [de-ergative--SG]
prefix can be considered a general object marker or an object
agreement marker...[4] the patient associated with a
de-ergativized verb does not appear in an oblique case, and
retains all other syntactic properties of transitive patients.

However, among the functions which Franchetto lists for the de-ergative
prefix is that of forming object relative clauses -- clearly retained from
the etymological funtion. She gives one example, reproduced here as 9. The
relative clause appears at the end of the main clause, separated from the
noun it modifies by the rest of the sentence. The verb ame 'take' bears the
the prefix g- ([N], cognate to the *nɨ- prefixes we have seen elsewhere), the
perfective nominalizer suffix -pârâ, and the plural suffix -ko. The second
person notional A possesses the derived nominal by means of the prefix a-
‘2’. Franchetto points out that 9 “could be translated literally as ‘you
will bring your taken water’.”

4
(9) tugá igi-nâm-ígo e-heke-ni a-g-ame-pârâ-ko
water bring-Punctual-Future 2-Erg-Plural 2-DeErg-take-Perfect-Plural
‘You will bring the water that you will take.’ (F’s 11a, p. 413)

Franchetto further describes how synchronically a “de-ergativized” verb is

required in all pragmatically marked constructions in which the direct object
is questioned, and also how it functions to mark main clause ‘interactive
moods’. Futher research is needed to understand the motivation for extending
the nï- object nominalizer into all of these functional areas.

For Makushi, Carson (1982:126) lists nɨ- as the object relative pronoun in a
restrictive relative clause, but her examples corroborate Abbott’s (1991:95)
analysis of nî- as an object nominalizer. In 10, Carson shows the verb eramá
‘see’ bearing the n- prefix and the -pɨ ‘Perfective’ suffix, then being
possessed by the first person notional A via the prefix u- ‘1’. In 11a-b,
Abbott shows the n- prefix occurring on the verb tîrî ‘give’, with the zero
suffix in 11a and with the past tense suffix -'pî in 11b. In both 11a-b, the
first person notional A possesses the derived nominal.

(10) pemóNkóN u-n-eramá-pɨ wutɨɨ-pɨ wɨt-tá

man 1-Rel-see-Perfective go-Perfective home-Directional
'The man whom I saw went home.' (Carson, p. 126)

(11) a. u-n-tîrî b. u-n-tîrî-’pî

1-O.Nomlzr-give 1-O.Nomlzr-give-Past
‘what I give’ ‘what I gave’ (A’s 371a-b. p. 95)

In Panare, like in Kuikúro, verbs bearing the n- prefix have been reanalyzed
for use as verbs in main clauses. In Gildea (1991) I argue that this
reanalysis has allowed an analogical extension of the Set I tense suffixes to
co-occur with the nɨ- prefix even when it serves its nominalizing function
(as in 12b). Thus far, Panare appears to be the only Cariban language where
a modern descendent of the *nɨ- prefix can co-occur with Set I tense
suffixes. Even so, the morphosyntax associated with the nɨ- prefix in Panare
is the same as that already described for the other languages: in 12a, the
verb ɨkɨtë ‘cut’ bears the nɨ- prefix, the perfect inferential nominalizing
suffix -jpë, and is possessed by second person a-. In 12b, the nɨ- prefix
occurs on the verb petyúma ‘hit’, which also bears the Set I past tense
suffix -yaj. This derived noun then bears the nominal animacy-marking suffix
-nëj and is possessed by the second person free pronoun amën.

(12) a. a-n-ɨkɨtë-jpë
2-O.Nmlzr-cut-Perfect.Inferential
‘the one that you cut’

b. amën nïpétyumayaanëj
amën Ø-^-n-petyúma-yaj-nëj
2Sg Ø-Free.Possessor-O.Nmlzr-hit-T/A-Animate
‘the one who you hit’ (lit. ‘your hittee’)

Armellada (1943:125) describes a n- prefix as a means of forming object

relative clauses, and, like Carson (1982) for Makushi, he considers the n- to
be a relative pronoun. In 13 the verb amue ‘roof’ bears the n- prefix, the
Set II past tense suffix -pue, and is possessed by the first person prefix u-
. The notional O tapui ‘house’ also occurs, as a free noun in apposition to
the derived nominal (with which it is coreferential).

5
(13) tapui u-n-amue-pue
house 1-O.Nmlzr-roof-Past
‘The house that I roofed.’ (A, p. 125)

For Tiriyo, Leavitt (1971) gives several paradigms of what he calls relative
clauses, showing how the addition of the nɨ- prefix changes them from being
possessed by the notional O to being possessed by the notional A. From those
paradigms, I have extracted two with the verb ekarama ‘give’. In 14a, the
verb bears the n- prefix with no suffix, and the resultant derived nominal is
possessed by third person i-. In 14b, the verb bears the n- prefix, the past
tense nominalizer -hpë, and then is possessed by second person ë-.

(14) a. i-n-ekarama
3-O.Nomlzr-give
‘what he is giving’ (Leavitt 1971)

b. ë-n-ekarama-hpë
2-O.Nmlzer-give-Past.Nmlzr
‘What you gave’ (Leavitt 1971)

For Wayana, both Jackson (1972) and de Goeje (1946) give examples of the *nɨ-
prefix co-occurring with a zero suffix, which each elsewhere describes as a
finite past tense inflection (Gildea 1992:97). However, when nɨ- co-occurs
with a zero suffix, there is no sense of past time, but rather the glosses
reveal a present continuous or habitual meaning. This corresponds to the
meaning of the zero suffix which is an action nominalizer in other Cariban
languages (Gildea 1992:121ff), so perhaps in Wayana both the Proto-Cariban *-
Ø ‘Action Nominalizer’ and *-Ø ‘Past’ have survived. In 15a, the verb epa
‘teach’ bears the n- prefix with a zero suffix, and is possessed by ï- ‘1’.
In 15b, the verb ïlï ‘make’ bears the n- prefix, the past tense nominalizing
suffix -tpï, the plural marker -kom, and is possessed by ku- ‘1+2’. In 16,
de Goeye (1946) shows the verb ili ‘make’ bearing the n- prefix with a zero
suffix, possessed by ë- ‘2’. The derived noun is the predicate noun of a
predicate nominal clause, with the subject being the demonstrative pronoun
sïn ‘this’.

(15) a. ï-n-epa
1-O.Nmlzr-teach
‘the one that I am teaching.’ (J, p. 58)

b. ku-n-ïlï-tpï-kom
1+2-O.Nmlzr-make-Past.Nmlzr-Plural
‘the thing we made; our former making’ (J, p. 59)

(16) ë-n-ili sin?

2-O.Nmlzr-make this
‘Is this your work?’ (de G, p. 109)

For Waiwai, Hawkins (1991) illustrates the *nɨ- prefix via the clause in 17:
the verb ekî ‘bring’ bears the prefix ñ- and the past tense nominalizer -tho,
then is possessed by the preceding noun paranakarî ‘white man’.

6
(17) krapa weeñasî paranakarî ñekîtho
krapa we-eña-sî paranakarî Ø-ñ-ekî-tho
bow 1-see-T/A white.man Ø-O.nmlzr-bring-Past.Nmlzr
‘Let me see the bow which the white man brought.’ (H, 1991b)

Finally, Adam (1893:51) lists three more languages with the *nɨ- prefix under
the heading “Prefixation of Two Personal Indices”: Chayma, Cumanagote, and
Tamanaco. Adam takes the notional A possessor a the first personal indice,
then the *nɨ- prefix as an indice of O. For Chayma, examples 18a-b show the
verb acarama ‘inform’ bearing both the n- ‘O Nominalizer’ prefix and the -r
‘possessive/nominalizer’ suffix, and being possessed by the notional A --
first person in 18a and second person in 18b. For Cumanagote, examples 19a-b
show the verb stem apchama ‘crush’ bearing the n- prefix, the -r suffix, and
then the locative yau; the notional A possessor is second person in 19a and
third person in 19b. For Tamanaco, example 20 shows the verb ene ‘see’
bearing the n- prefix, the nominalizing suffix -ri, and possessed by the
second person prefix a-.

(18) Chayma

a. u-n-acarama-r
1-O.Nmlzr-inform-Nominalizer
‘I inform him (about it).’

b. a-n-acarama-r
2-O.Nmlzr-inform-Nominalizer
‘You inform him (about it).’

(19) The Cumanagote n- ‘Object Nominalizer’

a. a-n-apchama-r-yau
2-O.Nomlzr-crush-Nominalizer-Locative
‘When you crush it...’ (A, p. 51)
(lit. ‘at it becoming your crushed thing’)

b. i-n-apchama-r-yau
3-O.Nomlzr-crush-Nominalizer-Locative
‘When he crushes it...’ (A, p. 51)
(lit. ‘at it becoming his crushed thing’)

(20) Tamanaco

a-n-ene-ri
2-O.Nmlzr-see-Possessed
‘that which you see’ (Adam 1893:51)
(lit. ‘your seen one’)

This concludes the exposition of the comparative Cariban data. In §4 I

discuss the conclusions which can be drawn from this data.

4 Conclusions: Towards Reconstruction and Classification

Before addressing the question of whether the Proto-Cariban *nɨ- is cognate
to the Proto-Tupí-Guaraní *emi- we must first determine that Proto-Cariban
*n- was, in fact, an object nominalizer. Franchetto (1990) asserts that the
Kuikuro ‘De-ergative’ prefix g-/ñ- is cognate to a third person verbal prefix
in Apalaí, Galibi, Hixkayana, and Waiwai, and hence represents the first
evolutionary step from a historically prior Proto-Cariban ergative verbal
system to a new, innovative split-nominative system (instantiated in the

7
verbal person-marking paradigm of the other four Cariban languages).
However, the data in §3 of this paper make it clear that the Kuikuro de-
ergative prefix is cognate to an idiosyncratic O nominalizer in all other
Cariban languages for which there is evidence. In contrast, the third person
prefixes in the other four Cariban languages are descended from an apparently
unrelated Proto-Cariban third person verbal prefix *n(ɨ)- (Gildea
1992:86ff).5

Although her comparative claims do not stand, Franchetto’s analysis of the

de-ergative construction as a recent innovation in Kuikúro appears completely
sound. A parallel innovation is independently attested in Panare (Gildea
1991) and possibly has happened also in Chayma (cf. example 18 above, where
Adam (1893) translates the erstwhile nominalization as a finite clause).

Given that it reconstructs to Proto-Cariban as an O nominalizer, we can now

compare the Proto-Cariban form *nɨ- with Proto-Tupí-Guaraní *emi- as forms
with similar morphosyntax and meaning. The question then becomes whether
they are in fact cognate, or whether they might both have been innovated by a
similar mechanism. As yet, the comparative evidence is insufficient to argue
convincingly for cognacy.

First, the forms are not transparently related phonologically: *emi- has two
syllables, a bilabial nasal, and a front vowel; *nɨ- one syllable, an
alveolar nasal, and either a central or back vowel. In order to argue for
cognacy, we must be able to tell a convincing story about the phonological
changes that each form has gone through in evolving away from their common
source form. Without much more comparative reconstruction within the Tupían
stock, such a story cannot be told.

Ordinarily, the isomorphism of the dependent clause systems alone would argue
for close relationship (cf. Derbyshire, this volume). Within that context,
the shared uniqueness of the morphosyntax associated with the respective
object nominalizers would constitute a further, quite strong argument for
cognacy. However, as pointed out by Aikenvald (pc) and Doris Payne (pc), the
isomorphism in Cariban and Tupí-Guaraní subordinate clause morphosyntax is
part of a wider areal pattern, shared by Arawakan, Gê, and possibly Panoan
languages as well. Thus, such patterns cannot be used to argue for an
especially close relationship between the Cariban and Tupí-Guaraní families.
Given that the identity in subordinate clause morphosyntactic patterns could
be inherited from a broader areal pattern, it is entirely plausible that two
of the families in the area could have independently innovated similar
exceptions to that pattern.

In line with this thought, Jensen (1990:127-8) suggests that the *emi
construction appears very similar to noun-incorporation (which is productive
elsewhere in the Tupí-Guaraní family). In particular, this would explain the
otherwise unattested pattern of the notional A possessing the derived
nominal. That is, if the nbotional O is represented as an incorporated
pronoun (which derives an intransitive verb), then the notional A becomes
simply an S; when this derived intransitive verb is nominalized, there is
nothing unique about its being possessed by the S. If a free pronoun *emi
can be reconstructed for the Tupían Stock (or even within the Tupí-Guaraní
family), and if there is evidence that pronouns can be incorporated, then the
nominalizing prefix in Tupí-Guaraní would perhaps be a more recent
innovation, too recent to be cognate with the Proto-Cariban prefix. Possibly
a parallel explanation might be found for the Proto-Cariban *nɨ- (cf.
speculation in Gildea to appear).

8
In conclusion, this paper offers a solid reconstruction of the Proto-Cariban
O nominalizing prefix *nɨ-, notes the similarity in both phonology and
morphosyntax between this form and the Proto-Tupí-Guaraní O nominalizing
prefix *emi-, but cannot offer solid arguments why the two should be
considered cognate.

Notes

* Support for research and writing this paper provided in part by The
Wenner-Gren Foundation, the American Philosophical Society, the University of
Oregon Doctoral Research Fellowship, and NSF grant No. DB5-9210130. An
earlier version of this paper was presented at the 47th Congress of
Americanists, New Orleans, July 1991. In revising, I benefitted from the
active discussion there, particularly from comments by Alexandra Aikenvald,
Desmond Derbyshire, Wolf Dietrich, and Doris Payne -- none of whom
necessarily agree with my conclusions here. All mistakes are my own.

1 I use the terms A, S, and O following Dixon 1979: A =

subject of a transitive clause, S = subject of an intransitive
clause, and O = direct object of a transitive clause. These are
meant as cover terms for syntactic relations, not semantic case
roles (i.e. A is NOT always a semantic agent.

2 Although Gildea (1992) reconstructs this prefix as a nominalizer, Hoff

(in personal correspondence) has suggested that in Carib of Surinam (and
perhaps in other languages) this prefix is not itself a nominalizer, but
rather it always co-occurs with a nominalizing suffix. By Hoff’s analysis,
Carib ni- has only the personal function of giving the genitive slot to the
A. In Carib, it is important to note that the derived nominal which bears
ni- does not always refer to the notional O of the nominalized verb, but can
sometimes refer to the action itself. Where this is true, there is little
point in linking n- to O at all. However, in most Cariban languages, the
addition of n- to a derived nominal changes the meaning such that the derived
nominal refers to the notional O; in at least two languages, Panare and
Wayana, the n- prefix nominalizes a verb which bears no other nominalizing
morphology. As such, I maintain my use of the term “O nominalizer”.

3 Koehn and Koehn’s orthography differs from the others presented in this
paper: y = [ï] and h = ['].

4 While other linguists have suggested that cognates to Kuikuro g-/ñ- be

thought of as generalized object markers (Williams 1932 and Carson 1982 each
analyze the cognate form in Makushi as an object relative pronoun, as does
Armellada 1943a for the cognate in Pemóng), Franchetto’s use of the term
“object agreement marker” seems unusual in that is there is no sense in which
this prefix agrees with any semantic features of the notional object -- it is
invariant regardless of person, number, gender, etc.

5 Although cf. Gildea (to appear) for a possible connection between the
two at the level of pre-Proto-Cariban.

References

9
Abbott, Miriam. 1991. “Macushi”. Handbook of Amazonian Languages, vol. 3.
Desmond C. Derbyshire and Geoffrey K. Pullum (eds), 23-1??. Berlin:
Mouton de Gruyter.
Adam, Lucien. 1893. Matériaux pour servir à l’établissement d’une grammaire
comparée des dialectes de la famille caribe. Bibliotèque Linguistique
Américaine, vol. 17.
Armellada, Cesário de. 1943. Grammatica de la Lengua Pemón, vol. I.
Caracas: Artes Gráficas.
Carson, Neusa. 1982. Phonology and Morphosyntax of Macuxi (Carib).
University of Kansas, Ph.D. dissertation.
Derbyshire, Desmond C. 1985. Hixkaryana and Linguistic Typology.
Arlington: Summer Institute of Linguistics and University of Texas at
Arlington.
Derbyshire, Desmond C. This volume. “Clause subordination and
nominalization in Tupi-Guaranian and Cariban languages.”
Dixon, R.M.W. 1979. “Ergativity”. Language 55.59-138.
Franchetto, Bruna. 1990. “Ergativity and nominativity in Kuikúro and other
Carib languages.” In Amazonian Linguistics: Studies in Lowland South
American Languages, Doris Payne (ed), 407-28. Austin: U of Texas
Press.
Gildea, Spike. 1991. “From object nominalization to subject focus
construction in Panare”. Presented at the Summer Meeting of Society
for the Study of Indigenous Languages of the Americas, Santa Cruz, June
1991
Gildea, Spike. 1992. Comparative Cariban Morphosyntax: On the Genesis of
Ergativity in Independent Clauses. Unpublished Ph.D. Dissertation,
University of Oregon.
Gildea, Spike. to appear. “Semantic and pragmatic inverse -- “inverse
alignment” and “inverse voice” -- in Carib of Surinam”. The Pragmatics
of Voice: Active, Inverse, Passive. T. Givón (ed). Amsterdam: John
Benjamins.
de Goeje, Claudius H. 1946. Études Linguistiques Caraïbes, v. 2. Amsterdam:
Verhandelingen van de Koninklijke Akademie van Wetenschappen, Afdeeling
Letterkunde, n.r. 49 (2).
Greenberg, Joseph. 1987. Language in the Americas. Stanford: University of
Stanford Press.
Hall, Katherine. 1988. The Morphosyntax of Discourse in De'kwana Carib.
Washington University in St. Louis, Ph. D. dissertation.
Hawkins, Robert. 1991. An analysis of the Wai Wai language. ms.
Hoff, Berend J. 1968. The Carib Language. The Hague: Martinus Nijhoff.
Jackson, Walter S. 1972. “A Wayana grammar”. In Languages of the Guianas,
Joseph E. Grimes (ed), 47-77. Norman: Summer Institute of Linguistics
and University of Oklahoma Press.
Jensen, Cheryl. 1990. “Cross-referencing changes in some Tupí-Guaraní
languages”. In Amazonian Linguistics: Studies in Lowland South
American Languages, Doris L. Payne (ed), 117-58. Austin: U. of Texas
Press.
Koehn, Edward and Sally Sharp Koehn. 1986. “Apalai”. In Handbook of
Amazonian Languages. Desmond C. Derbyshire and Geoffrey K. Pullum
(eds), 33-127. Berlin: Mouton de Gruyter.
Leavitt, Claude W. 1971. Trio Grammar. ms.
Loukotka, Cestmir. 1968. Classification of South American Indian Languages.
Los Angeles: Latin American Center, UCLA.
Rodrigues, Aryon D. 1985a. “Evidence for Tupí-Carib relationships.” In
South American Indian Languages: Retrospect and Prospect. Harriet E.
Manelis Klein and Louisa R. Stark (eds), 371-404. Austin: University
of Texas Press.

10
Williams, James E. 1932. Grammar Notes and Vocabulary of the Language of
the Makuchi Indians of Guiana. St. Gabriel Mödling near Vienna,
Austria: Verlag der Internationalen Zeitschrift “Anthropos”.

11