Adelaar Willem, Muysken Pieter. The languages of the Andes

Подождите немного. Документ загружается.

264 3 The Inca Sphere

kingdom of the Lupaca, centred around the town of Chucuito, southwest of Lake

Titicaca. It is relatively well known thanks to Garci Diez de San Miguel’s report of 1567

(see section 3.3.1), which contains valuable data about the organisation of an Aymara

chieftaincy and its colonies in the coastal region. Shortly after the arrival in 1568 of the

ﬁrst Jesuits in Peru, a mission of that Order was established at Juli in the Lupaca area.

Through the work of Bertonio, among others, the Lupaca dialect of Aymara became

representative of the language as a whole.

3.3.3 Internal variation in the Aymaran language family

As we have seen in section 3.1, the northern and southern branches of the Aymaran

linguistic family are separated by a considerable geographical distance. This separation

is not only geographical, it is also a matter of demography. The demographic factor makes

the comparison between the Aymara language, on one hand, and its two small northern

relatives Jaqaru and Cauqui, on the other, a rather unbalanced one. As the incipient

dialectological studies progress, new differentiating elements may emerge within the

Aymara domain itself. So far, a moderate amount of internal variation in terms of dialects

has been detected (Briggs 1993).

Jaqaru and Cauqui are characterised by unusually rich obstruent inventories, which

are identical for both varieties. The Aymara obstruent inventory, although relatively

complex, is less elaborate. All three languages distinguish between plain, aspirated and

glottalised obstruents. But, whereas Aymara has ﬁve different articulations (bilabial,

alveolar, velar and uvular stops; palatal affricates), Jaqaru and Cauqui have eight (the

abovementioned ﬁve; alveopalatal stops; alveolar and retroﬂex affricates). Furthermore,

Jaqaru and Cauqui have two sibilants (alveolar and palatal), whereas Aymara has only

one (alveolar). The status of the additional obstruent series in Jaqaru and in Cauqui,

either as cases of conservatism or as innovations, has been a topic of debate. As it stands

now, only the retroﬂex affricate series seems to bear out its conservative nature. The

velar nasal has phonemic status in Jaqaru, in Cauqui, and in a very limited number of

Aymara dialects (see below).

Most observations concur in suggesting that Jaqaru and Cauqui are mutually intelli-

gible to a great extent, and that the extent of the lexical, morphological and phonological

differences existing between them is limited (see section 3.1). Nevertheless, the distinct

nature of the two varieties is highlighted by Hardman (1975, 1978). She points, among

other things, at an innovative process of vowel harmonisation in sufﬁxes that sets Jaqaru

apart from both Cauqui and Aymara. It is illustrated by object-marked verb forms such

as Jaqaru irp-k-utu ‘he takes me along’, in contrast with Cauqui irp-k-itu

and Aymara

The use of an ongoing event marker -k- is required for present tense in the central Peruvian

members of the Aymaran family.

3.3 The Aymaran language family 265

irp-itu (Hardman 1975: 440). Another example of this vowel harmony can be found

in the Jaqaru verbal nominalisation afﬁx -nuˇsu, e.g. in irp-nuˇsu ‘to take along’, which

corresponds to Cauqui irp-niˇsu.

In her dictionary of the Jaqaru language, Belleza Castro (1995: 55–6) observes that the

alveolar affricate series of Jaqaru is regularly matched by the retroﬂex series in Cauqui,

e.g. Jaqaru haca-, Cauqui haˇc

a- ‘to cry’ (cf. Aymara haˇca-); Jaqaru c’a:ka, Cauqui

ˇc

’a:ka ‘bone’ (cf. Aymara ˇc’aka, ˇc’ak

a). It does not mean, however, that the alveolar

affricate series is entirely absent from Cauqui (e.g. Jaqaru, Cauqui ac’iki ‘cold’). A

lexical feature of Cauqui is the use of a deictic pronoun uwa ‘that’, where both Aymara

and Jaqaru have uka (Belleza Castro 1995: 185). Unfortunately, only incidental data

are available for Cauqui. A substantial vocabulary and a grammatical overview of the

language would be needed in order to appreciate the correspondences. As long as this

condition is not met, any reconstruction of Proto-Aymaran will remain tentative.

Present-day dialectal variation in the Aymara language area has been studied in a

comprehensive manner by Briggs (1976, 1993); for a critical assessment of her work

see Cerr´on-Palomino (1995a). In-depth studies of individual Aymara dialects are scarce,

except for the variety spoken in and around the Bolivian governmental residence city

La Paz. The Aymara of the department of La Paz in Bolivia lies at the basis of the most

authoritative and comprehensive studies of that language, such as Hardman et al. (1974,

1988), Herrero et al. (1978) and Yapita (1991). Porterie-Guti´errez (1988) deals with

Peruvian varieties (that of Chucuito, in particular), and some work has been published

on the Chilean variety (Clair-Vassiliadis 1976; Poblete and Salas 1997; Salas and Poblete

1997). Dictionaries tend to combine lexical items from different localities, although the

selection is usually limited by national boundaries.

Briggs’s work contains a wealth of data concerning variation in the Aymara language

at all grammatical levels. On the basis of her ﬁndings, she proposes two classiﬁcations

of the geographic varieties of Aymara (Briggs 1993: 388–98). The ﬁrst classiﬁcation

distinguishes between northern Aymara (roughly coinciding with the varieties of the de-

partment of Puno in Peru and La Paz in Bolivia), southern Aymara (in the departments of

Oruro and Potos´ı, Bolivia, and in northern Chile), and an intermediate group consisting

of the dialects spoken in the Peruvian departments of Moquegua and Tacna on the south-

western Paciﬁc slopes. These dialects are said to have characteristics in common with

both the northern and the southern group, there being particularly important similarities

between the dialect spoken in the interior of Tacna (Tarata) and that of the Carangas area

in western Oruro (Briggs 1993: 401). The second classiﬁcation proposed establishes a

contrast between a central (supposedly innovative) dialect group in the area surrounding

the city of La Paz and a peripheral dialect group comprising all the outlying areas.

Many cases of variation registered by Briggs seem to concern individual lexical items

and morphemes, rather than to reﬂect regular changes. Some cases, however, do indicate

266 3 The Inca Sphere

systematic developments, such as the voicing of plain stops after nasals and laterals in

Salinas de Garci Mendoza in southern Oruro, for instance, in tunga ‘ten’ (< tunka),

ambara ‘hand’ (< ampara), and qal

-da- ‘to begin’ (< qal

-ta-); cf. Cerr´on-Palomino

(1995a: 122).

As has been anticipated, there is one case of dialect variation that, in particular,

deserves the attention of historical comparative linguists, that is, the development of a

velar nasal in non-automatic environments.

At least three Aymara dialect areas, Tarata

(in Tacna, Peru), Carangas (in Oruro, Bolivia) and the Aymara-speaking part of northern

Chile have preserved a ﬁrst-person possessive and a ﬁrst-person future subject ending

in -

a, e.g. uta-

a ‘my house’, sar-x

a ‘I shall go (home)’. They share this feature

with Jaqaru and Cauqui, where the same sufﬁx is used. Both in Tarata Aymara and in

Jaqaru the sound

is also found intervocalically in a small number of lexical items (e.g.

Tarata a

anu ‘face’, ‘cheek’; Jaqaru i

aca ‘servant’). In spite of the limited phonotactic

possibilities of the distinctive velar nasal – it only occurs between vowels, of which the

second one may or may not be suppressed – there seems to be no reason not to reconstruct

it for Proto-Aymaran.

Already in the Lupaca variety described by Bertonio (1603a, b) most velar nasals had

been replaced by a velar fricative h. Other dialects eliminated the velar nasal with its

low phonemic load in different ways. The ﬁrst-person possessive and ﬁrst-person future

endings *-

a were replaced by elements such as -ha,-x

a,-n

a and/or vowel length, and

exhibit considerable dialectal variation at this time.

Aspirated and glottalised obstruents are widely used in the Aymaran languages. Both

categories are held to represent features inherited from the proto-language. However,

although the presence of aspiration or glottalisation is stable in many lexical items and

afﬁxes, it can be variable in others. Examples are the ﬁrst-person non-future subject

marker -t

a and its homophone, the ablative case marker -t

a. These sufﬁxes have lost

their aspiration in La Paz Aymara, whereas it has been retained in most other Aymara

dialects and in Jaqaru. The loss of aspiration brought about a formal coincidence of these

elements with the second-person non-future subject marker -ta and the nominaliser -ta,

but is, at the same time, responsible for the different morphophonemic behaviour of

An automatic environment would be the position before k within a root (e.g. in tunka ‘ten’),

where the nasal is velar by assimilation.

Cerr´on-Palomino (1994a: 111; 1995a: 114–17) observes that the velar nasal sound is followed by

avelar fricative in several Aymara dialects (e.g. a

hanu ‘face’ in Guallatire, northern Chile; and a

ﬁrst-person future marker -

ha in Conima, Huancan´e, Peru). He also points at the correspondence

between Aymara manq

a [maŋq

a] ‘under’, ‘inside’ and Jaqaru ma

a ‘below’, and concludes

that the nasal velar in Aymaran must take its origin from a preconsonantal allophone of the plain

nasal n.However, in view of such pairs as Jaqaru ya

a and Quechua yana ‘companion’, and

Jaqaru ya

-iˇsi-, Quechua yana-pa- ‘help’, we are inclined to opt either for an inherited though

obsolescent distinction, or for a retraction of the alveolar nasal in intervocalic position before a.

3.3 The Aymaran language family 267

the two pairs of sufﬁxes. The aspiration, whether present or not, entails the loss of the

vowel -a in a sufﬁx before certain independent sufﬁxes,

such as -wa, e.g. mun-t-wa

{mun(a)-t

(a)-wa} ‘I want it’, but mun-ta-wa {mun(a)-ta-wa} ‘you want it’.

3.3.4 Salient features of the Aymaran language family

The Aymaran language family consists of languages which are structurally very similar

to Quechua. The similarities between the two groups are so obvious and so pervasive

that they can only be explained by a long period of shared history and complex mutual

relations. The structural similarities are not merely superﬁcial. They consist of highly

speciﬁc semantic and syntactic parallels that, if not due to common origin, must be the

result of long and intensive bilingual interaction. More than 20 per cent of the lexicon that

can be reconstructed for the proto-languages of both families is either identical, or nearly

so. The reconstructed phoneme inventory of Proto-Aymaran is also partly identical to

that of Proto-Quechuan, although there are differences which deserve close attention.

It is remarkable, under these circumstances, that the languages of the Aymaran family

have managed to preserve two characteristics that set them apart not only from Quechua,

but also from the other known languages of the Andean region. These characteristics

belong to the domain of phonotactics and morphophonemics. Aymaran roots, both verbal

and nominal, must end in a vowel. In order to meet this condition, borrowed nouns with

a ﬁnal consonant in the original language become Aymaranised by the addition of a ﬁnal

vowel, regardless of whether the language borrowed from is Quechua, Spanish, or any

other. This process is still productive today, and it has been in existence in the Aymaran

languages for as long as evidence is obtainable by projection into the past.

(119) Proto-Quechua *paˇc

ak ‘hundred’ Aymara pataka

Jaqaru paˇc

aka

Proto-Quechua *kuntur ‘condor’ Aymara kunturi

Jaqaru k´untiri

Spanish habas ‘beans’ Aymara hawasa

Jaqaru h´awaˇsa

The second characteristic concerns the allomorphic shape of roots and sufﬁxes. Like

Quechua, the Aymaran languages are predominantly agglutinative, using sufﬁxes almost

exclusively. Unlike Quechua, however, these sufﬁxes are accompanied by rules leading to

the suppression of a preceding vowel or, under given circumstances, of vowels belonging

to the sufﬁx itself. From a synchronic point of view, the vowel suppression rules are

seldom phonologically motivated. The following examples of derivation of the verb

For the notion of independent sufﬁxes see sections 3.2.6 and 3.3.6.

The Jaqaru forms are from Belleza Castro (1995).

268 3 The Inca Sphere

apa- ‘to bear’, ‘to carry’ illustrate the different behaviour of Aymara sufﬁxes with an

initial s. The sufﬁx -su- ‘outward motion’ must be preceded by vowel suppression, the

sufﬁx -si- ‘reﬂexive’ has no such effect:

(120) ap-su- ‘to take out’

apa-si- ‘to take one’s share’, ‘to take for oneself’

The corresponding sufﬁxes in Jaqaru are -ˇsu- and -iˇsi-, respectively. Both sufﬁxes

trigger vowel suppression.

(121) ap-ˇsu- ‘to take out’

ap-iˇsi- ‘to take one’s share’, ‘to take for oneself’

It would be tempting to seek an explanation for the different behaviour of Aymara -si-

and -su- in the Jaqaru facts. However, this does not seem to be possible. The presence

of a sufﬁx-initial vowel i in Jaqaru -iˇsi-isdue to innovative vowel harmony, which

could develop precisely in environments where the vowel was preserved at ﬁrst (*apa-

ˇsi- > ap-iˇsi-). Some cases of vowel suppression may eventually be explained by other

diachronic developments that are not yet understood. From a phonological point of view,

it is signiﬁcant that Aymara sufﬁxes with an initial l

or y are never preceded by vowel

suppression and with other resonants only exceptionally so (Briggs 1993: 55–6).

The Aymaran vowel suppressions can produce spectacular sequences of consonants,

not separated by vowels even at the phonetic level. This occurs when several sufﬁxes

triggering vowel suppression appear in a sequence. Consider the following example from

La Paz Aymara:

(122) hani-w hisk-t’-k-t-ti

{hani-w(a) hisk

(i)-t’(a)-k(a)-t(a)-ti}

not-AF ask-M-AN-1S-NE

‘I did not ask him.’

(Yapita 1991: 75)

The three sufﬁxes -t’a-, -ka- and -ta trigger the suppression of a preceding vowel.

Additionally, the ﬁrst-person subject sufﬁx -ta loses its own vowel before independent

sufﬁxes, such as -ti. This loss can be explained phonologically by the fact that -ta

originally began with an aspirated t

,asitstill does in many dialects.

The sufﬁx -ti

itself exerts no inﬂuence on the preceding vowel. The aspiration of k

in hisk

i-is

also lost in the process. The sufﬁx -wa loses its vowel because it marks the end of a

major preverbal constituent of the sentence. Although Aymara and Jaqaru differ in the

inventory of their sufﬁxes and the nature of the accompanying morphophonemic rules,

vowel suppression plays a central role in both languages.

Aspiration is sometimes associated with vowel loss in the Aymaran languages.

3.3 The Aymaran language family 269

Table 3.6 Jaqaru personal reference markers

Pronouns Possessive endings Imperative

1 pers. na -

a-V

2 pers. huma -ma -ma

3 pers. upa -p

a -pa

4 pers. hiwsa -sa -Vtna

In contrast, the structure of roots is relatively simple. Root-internal consonant clusters

may consist of two consonants at the most (although Jaqaru has some exceptions). Root-

initial clusters do not occur. In comparison to Quechua, Aymaran roots tend to contain

more open syllables. In Aymara the possibilities of clustering (that is, previous to the ap-

plication of any vowel suppression rules) are limited. When root-internal clusters contain

a stop, it is always the second member in the cluster. Clusters of stops and/or affricates

do not normally occur (Martin 1988: 44–7). Jaqaru is less restrictive in this respect.

In the domain of morphology, the Aymaran languages are characterised, among other

things, by a fourfold division of the category grammatical person, based on the inclusion

or non-inclusion of speaker and addressee, respectively (cf. Hardman et al. 1988: 18).

The system comprises four basic units: ﬁrst person (+ speaker, – addressee), second

person (+ addressee, – speaker), third person (– addressee, – speaker), and fourth person

(+ addressee, + speaker). The Aymaran fourth person coincides with what is known in

linguistic literature as an inclusive ﬁrst person plural. There are no speciﬁc endings for the

exclusive ﬁrst person plural, which, apart from the optional presence of plural marking,

are the same as the ﬁrst-person-singular endings. Each of the four units is represented

by speciﬁc underived morphemes, which surface in the shape of the personal pronouns,

in the nominal possessive endings and in parts of the verbal paradigm. Table 3.6 gives

a synopsis of the basic personal pronouns, possessive endings and verbal imperative

endings in Jaqaru (Hardman 1983a). (‘V’ indicates that the vowel preceding the sufﬁx

is preserved; otherwise, suppression of the preceding vowel is required.)

Number distinctions do not play a fundamental role in the Aymaran personal reference

system. The three basically singular pronouns (1–3) can be pluralised by means of a

nominal plural marker: -naka in Aymara; -kuna (as in Quechua) in Jaqaru. In this way, a

secondary distinction is created between the inclusive fourth person and the (exclusive)

plural of the ﬁrst person. The pronoun ‘we (exclusive)’ is na-naka in Aymara and na-kuna

in Jaqaru; ‘we (inclusive)’ is hiwasa in Aymara and hiwsa in Jaqaru. Possessive endings

cannot be pluralised, except periphrastically. Verbal endings have no plural counterparts

either, but it is possible to express the notion of plurality internally in the verb form, as

in Pichacani (Puno) Aymara laru-si-px

-t

-wa (laugh-RF-PL-1S-AF) ‘we (excl.) laugh

at him.’

270 3 The Inca Sphere

3.3.5 Aymara phonology

The orthography currently in use for modern Aymara presents a number of distinct

elements, which have become widely accepted. It includes notations such as j for the

glottal (or velar) voiceless fricative, and x (sometimes also jj) for the uvular voiceless

fricative. Aspiration is either written as a double apostrophe (”) or as h.Inour discussion

of Aymara (and Jaqaru) we continue to use the symbols that we have used so far, namely,

h for the glottal (or velar) fricative, x

for the uvular fricative, raised

for aspiration and

raised

for palatalisation, in order to help preserve the unity of the presentation and

facilitate the comparison with other Andean languages, such as Quechua.

The vowel system of Aymara is trivocalic and consists of a low vowel a and two high

vowels, front unrounded i and back rounded u.AsinQuechua, the high vowels have mid

allophones in the environment of a uvular consonant.

Vowel length is distinctive. There are three long vowels, a:, i:, u:.Vowel length

plays a role in the morphology (see section 3.3.6). It can also, though seldom, be part

of the phonological makeup of root morphemes, in which case the vowel is almost

always a::

(123) ˇcaka ‘bridge’ ˇca:ka ‘stem of the quinua’

A frequent source of vowel length in Aymara is the optional suppression of y between

same vowels in the sequences aya and uyu.Variation of the type aya ∼ a:, uyu ∼ u: has

been recorded from the sixteenth century onwards. Nevertheless, the forms with vowel

length have not succeeded in replacing those with internal y,which are still predominant

today.

(124) contemporary Bertonio (1612a)

maya ∼ ma: maya ∼ maa ‘one’

tiyi ∼ ti: tiy ‘cave’

suyu ∼ su: suyu ∼ suu ‘parcel of land’, ‘share of work’

Long vowels are also found in compound expressions as a result of sandhi, when there

is a succession of vowels without an intervening consonant:

(125) hiˇc

u:ru ‘today’ from hiˇc

a ‘now’, uru ‘day’

The consonant inventory of Aymara is very similar to that of the Quechua II dialects

surrounding it (Cuzco Quechua, Bolivian Quechua). The consonant inventory of La Paz

Aymara is represented in table 3.7 (cf. Yapita 1991: xiv).

Both the existence of vowel length in Aymara and its use there are strongly reminiscent of the

Quechua I dialects. This is not the only respect in which Aymara resembles Quechua I, rather

than Quechua II (cf. Cerr´on-Palomino 1994a).

3.3 The Aymaran language family 271

Table 3.7 La Paz Aymara consonant inventory

Labial Alveolar Palatal Velar Uvular

Plain p t ˇckq

Obstruents Aspirated p

ˇc

Glottalised p’ t’ ˇc’ k’ q’

Fricatives s h x

Laterals l l

Nasals m n n

Vibrant r

Glides w y

Stress in Aymara is basically penultimate and has been described as non-phonemic

(Martin 1988: 43). Long vowels in word-ﬁnal position attract stress, and hence they are

sometimes interpreted as sequences of like vowels. Indeed, such cases are usually the

product of a reduction, as in (126):

(126) sar´a:

{sara-V: < *sara-

ŋa}

go-1S.F

‘I shall go’

Deviations from these basic rules are not uncommon. When a word ends in a consonant

or a sequence of consonants, this is normally the result of syntactic vowel suppression

(see below). In those cases stress is on the ﬁnal vowel. An exception is the second-person

imperative in La Paz Aymara, which ends in a consonant but has penultimate stress:

(127) s´ara-m

go-2S.IM

‘Go!’

The vowel suppression rules of La Paz Aymara can be divided into those that

precede and those that follow the assignment of stress. As we saw (section 3.3.4),

morphophonemic vowel suppression has a long history in the Aymaran languages. It

occurs at word-internal morpheme boundaries, and stress is assigned afterwards.

For

The form sara-

a is preserved in the Aymara-speaking areas of interior Tacna (Peru) and

Carangas (Bolivia).

Briggs (1993: 80–3) observes that the ﬁnal vowel of the second-person imperative ending -

m (<*-ma), which is normally absent, may be restored in some cases. It may lead to cases

of antepenultimate stress because the characteristic stress pattern of the imperative remains

unchanged (e.g. ap´a-ni-m ∼ ap´a-ni-ma ‘bring it here’).

Vowel suppression within the root is exceptional in Aymara. An example is wal

aqi- ‘to boil’,

which is reduced to wal

- in vowel suppressing derivations.

272 3 The Inca Sphere

instance, in (128) the ﬁrst root vowel is stressed, after the sufﬁx -t(a) ‘ﬁrst-person sub-

ject’ loses its vowel by suppression before the attenuating sufﬁx -x

a, the root ˇcura- ‘to

give’ loses its ﬁnal vowel by suppression before -t(a), and stress is assigned to the next

available vowel on the left. In (129) the second-person subject sufﬁx -ta does not lose

its vowel before -x

a, and, as it stands in penultimate position, it also bears the stress.

(128) ˇc´ur-t-x

{ˇcur(a)-t(a)-x

give-1S-TO

‘I gave it.’

(129) ˇcur-t´a-x

{ˇcur(a)-ta-x

give-2S-TO

‘You gave it.’

In contrast, syntactically motivated vowel suppression affects the ﬁnal vowel of any

major constituent of a sentence which does not occupy the ﬁnal position in that sentence.

The ﬁnal constituent, usually the verb, remains unaffected. Stress is assigned before

vowel suppression and, if present at all, it is positioned on the last vowel in the word

that is actually pronounced

(130) kun´a-t huk’amp-r´ak qul

qmun-x

-t´a-sti

{kuna-t(a) huk’amp(i)-Ø-rak(i) qul

q(i)-Ø mun(a)-x

(a)-ta-sti}

what-AB more-Z-AD money-Z want-CM-2S-CT

‘And how much more money will you need?’

(Yapita 1991: 99)

Words consisting of more than two syllables that occupy a non-ﬁnal position in a noun

phrase are also affected by the suppression of their ﬁnal vowel:

(131) ˇc’iy´ar ´uta

{ˇc’iyar(a) uta}

black house

‘a black house’

Aymara, furthermore, has a morphosyntactic rule of vowel suppression, which affects

the ﬁnal vowel of a nominal base (consisting of a bare noun, a noun followed by one or

The effect of the topic marker here has been characterised as ‘attenuation’ (Hardman et al. 1988:

280).

Aymara dialects may differ in the frequency and obligatory nature of syntactic vowel suppression.

Porterie-Guti´errez (1988: 46) qualiﬁes the loss of ﬁnal vowels in the dialect of Chucuito as ‘very

common’ (tr`es courant). In La Paz Aymara it is considered a standard practice (cf. Yapita 1991).

3.3 The Aymaran language family 273

more nominal afﬁxes, or a nominalised verb) and identiﬁes it as the direct object of a

transitive verb or the geographic goal of a motion verb. In the literature, these forms are

referred to as ‘zero complements’ (e.g. Briggs 1993: 141–3). Zero complements nor-

mally precede the verb. They can be followed by independent sufﬁxes, which themselves

lose their vowel due to the syntactic vowel suppression mentioned above:

(132) k

it-s suy-paˇca

it(i)-Ø-s(a) suy(a)-paˇca}

who-Z-IR wait-DP.3S

‘He must be waiting for someone?’

(Yapita 1991: 51)

(133) uk sar-ta

{uk(a)-Ø sar(a)-ta}

that place-Z go-1S

‘I went there.’

(Yapita 1991: 35)

The following sentence exempliﬁes the use of k

iti ‘who’ in the subject role, where

there is no zero complement vowel suppression:

(134) k

iti-s uta-r sara-ni

iti-s(a) uta-r(u) sara-ni}

who-IR house-AL go-F.3S

‘Who will go to the house?’

(Yapita 1991: 42)

It is often not possible to distinguish between zero-complement vowel suppression

and syntactically based vowel suppression, in particular when dealing with heads of noun

phrases not followed by independent sufﬁxes.

The word qul

q(i) in (130) is a case in

point. As can be seen from this example, as well as from k

it-s in (132), vowel suppres-

sion, both syntactically motivated and in its function as a zero complement, can produce

consonant clusters (including sequences of same consonants) in word-ﬁnal position.

Cerr´on-Palomino (1994a: 58–9) situates the issue of the zero complement in Aymara

within the context of a more general rule deleting the ﬁnal vowel of preverbal com-

plements.

In that perspective, the zero complement could be interpreted as a noun

unmarked for case which indicates a direct object or goal. It is true that the case markers

-na ‘genitive-locative’ and -ta/-t

a ‘ablative’ lose their vowel before any independent

The literature is silent about the relation between zero-complement vowel suppression and stress

assignment. Nevertheless, there is a general rule to the effect that in a (full) word form stress is

on the penultimate vowel but remains on that vowel when the ﬁnal vowel is suppressed. There are

exceptions, such as the second-person imperative (e.g. s´ara-m ‘go!’), but the zero complement

is not among those exceptions.

Briggs (1993: 142) points at the occurrence of postverbal zero complements in some varieties

of Aymara. This should lead to a different formulation of the rule in question.