Adelaar Willem, Muysken Pieter. The languages of the Andes

Подождите немного. Документ загружается.

194 3 The Inca Sphere

State-of-the-art books are scarce in the ﬁeld of Quechua studies. One book intended to

serve that purpose is B¨uttner (1983). It contains a good survey of older work on Quechua.

Cerr´on-Palomino’s Ling¨u´ıstica quechua (1987a) is the reﬂection of years of academic

teaching experience in almost every aspect of Quechua studies. It is particularly useful

for its insightful and equitable treatment of the existing literature. See also Cerr´on-

Palomino’s extensive survey of Quechua studies in Revista Andina (Cerr´on-Palomino

1985). Books dealing with historical developments surrounding the Quechua language

are Torero (1974) and Mannheim (1991). The latter work describes the fate of the Cuzco

dialect in colonial and republican times.

The 1980s brought a growing interest in the social factors that can contribute to the

survival of the Quechua language. Policies of standardisation, whether successful or not,

and several ambitious projects in bilingual education have given a renewed impulse to

Quechua studies. An example of a study dealing with the Puno experimental project for

bilingual education and its effects upon language maintenance by Quechua speakers is

Hornberger (1988).

At the time of our writing, there are no journals speciﬁcally dedicated to Quechua

studies, as there are none dealing with Andean languages in general. The journal Papers

in Andean Linguistics (1972–5) has remained a short-lived experiment. Articles on

Quechua have appeared, inter alia,inAllpanchis (Cuzco), Amerindia (Paris), Bulletin

de l’Institut Fran¸cais d’Etudes Andines (Lima, Paris), Indiana (Berlin), International

Journal of American Linguistics (Chicago), Latin American Indian Literatures Journal

(Pennsylvania), Lexis (Lima) and Revista Andina (Cuzco). The absence of specialised

journals is somewhat compensated by the frequent appearance of compilation works on

Quechua and on the central Andean languages in general (e.g. Cerr´on-Palomino 1982;

L´opez 1988; Cerr´on-Palomino and Sol´ıs 1990; Cole, Hermon and Mart´ın 1994).

Foranintroductory survey of texts and literary production in Quechua see

section 3.2.11.

3.2.5 Phonology

Foranevenly balanced treatment of Quechua phonology, it is best to take the recon-

structed phoneme inventory of Proto-Quechua as a starting-point. Roughly speaking,

In our discussion of Quechua language data the following orthographic conventions are used.

Mid vowels are written i, i:, u, u: when adjacent to a uvular consonant. The symbol q represents

any uvular consonant, although its realisation may vary according to the dialect exempliﬁed. The

possibilities that obtain are a voiceless or voiced stop, and a voiceless or voiced fricative. There is

no phonemic contrast between (plain) uvulars, except in Cuzco Quechua, where a voiceless stop

and fricative are contrastive between vowels, as in waqay ‘to cry’ versus wax

ay (< *waqyay ‘to

call’). The symbol n is used both for alveolar and for velar nasal allophones. The former occur

before vowels, the latter in word-ﬁnal position and before resonants. Before other consonants,

articulatory assimilation is the rule. Phonemic contrast between alveolar and velar nasals (as in

Chachapoyas and in northern Jun´ın/Tarma) is peripheral or exceptional.

3.2 The Quechuan language family 195

the phonological characteristics of the modern dialects can be derived from the Proto-

Quechua system by assuming regular sound change. A notorious exception is the exis-

tence of a contrast between plain, glottalised and aspirated stops and affricates in Cuzco

and Bolivian Quechua, and of a similar contrast between plain and aspirated stops in

highland Ecuadorian Quechua.

Orr and Longacre (1968) assumed that the three-way distinction between plain, glot-

talised and aspirated consonants had been inherited from an alleged Proto-Quechumaran

through the intermediate stage of Proto-Quechua. Within the Quechuan family, how-

ever, this distinction is limited to the subset of the QIIC dialects adjacent to the Aymara

linguistic area. Lexical correspondences are frequently inconsistent with respect to glot-

talisation and aspiration between dialects and even internally, within the same dialect

(e.g. Cuzco Quechua riku- ‘to see’ but rik

u-ri-‘to appear’, both from the same root riku-,

against Arequipa Quechua rik

u-, rik

uri-). Likewise, it has been shown that the corre-

spondences between Ecuadorian aspirated stops, on the one hand, and Cuzco–Bolivian

glottalised and aspirated stops on the other, are not systematic (Torero 1984). As a result,

the presence of glottalised and aspirated stops and affricates in Cuzco–Bolivian is often

seen as the effect of language contact between these dialects and Aymara, which has the

same contrasts. The expansion of glottalisation and aspiration into the native Quechua

lexicon was partly explained by a mechanism of iconicity elaborated in Mannheim and

Newﬁeld (1982); see also Mannheim (1991: 177–207). Ecuadorian aspiration has been

interpreted as a case of Cuzco adstratum, dating from the short period of Inca occupation,

possibly in combination with a putative legacy of the area’s pre-Quechuan languages.

There is no clear evidence that glottalised and aspirated stops and affricates belonged to

the Proto-Quechua phoneme inventory.

Reconstructions of the Proto-Quechua sound system are given in Torero (1964) and in

Parker (1969b). The differences between the two reconstructions reside in the treatment

of liquids, vibrants and long vowels. Parker reconstructs the consonants *l and *ˇr,

whereas Torero does not. On the other hand, Torero reconstructs vowel length, whereas

Parker only posits short vowels. As it appears now, *l and vowel length were marginally

present in Proto-Quechua; *ˇr wasanallophone of *r,which later developed into a

distinctive phoneme. It is interesting to observe that the palatal lateral *l

is a frequent

speech sound, whereas its plain alveolar counterpart l remains marginal in most dialects.

Proto-Quechua has a three-vowel system, consisting of an unrounded front vowel

i,arounded back vowel u and a low central vowel a.Ifalength distinction did exist

at all, it must have been marginal. The phonetic realisation of the vowels i and u was

variable. When adjacent to the uvular consonant q, they were pronounced as mid vowels

[e], [o]. They were high vowels [i], [u] elsewhere. Given the situation in the modern

dialects, it is likely that this allophonic lowering also occurred before clusters consisting

of a resonant (l

, n, r) and q, e.g. Cuzco Quechua sunqu [sɔŋqo] ‘heart’; pirqa [pεrqa]

‘wall’.

196 3 The Inca Sphere

Table 3.1 Proto-Quechua consonants

Labial Alveolar Palatal Retroﬂex Velar Uvular

Voiceless Stops p t k q

Affricates ˇcˇc

Fricatives s ˇsh

Voiced Nasals m n n

Laterals (l) l

Vibrants r (ˇr)

Glides w y

Table 3.2 Proto-Quechua vowels

Back Central Front

Short Long Short Long Short Long

High u (u:) i (i:)

Low a (a:)

The Proto-Quechua vowels and their rather wide range of allophonic realisations

have been preserved in a majority of the present-day dialects. At the same time, close

contacts with the dominant Spanish language led to the introduction of mid vowels

(e, o)incases where the phonological environment does not predict it. For present-

day bilingual speakers, Quechua has a ﬁve-vowel system.

Although it is true that in

Quechua borrowings from Spanish the mid vowels are replaced by high vowels in root-

ﬁnal position (e.g. nasi-, na:si- ‘to be born’ from Spanish nace ‘he/she is born’; platu,

pla:tu ‘plate’ from Spanish plato), they are often maintained elsewhere, as in mesa,

me:sa ‘table’ from Spanish mesa.

Dialects in which the uvular stop lost its uvular character almost without exception

have annulled the allophonic lowering rule. It illustrates the synchronic dynamism that

characterises the relation between the use of mid vowels and the presence of a uvular

consonant. The velar and uvular stops *k and *q have merged into k in all Quechua IIB,

e.g. Proto-Quechua *qul

qi [qɔl

qe] ‘money’, Ecuadorian (Chimborazo) Quechua kul

[kul

ki]. In QI Huanca the uvular became a velar fricative or a glottal stop, locally evolv-

ing into vowel length or zero, and the high vowel was then restored, e.g. Proto-Quechua

In literature and everyday imitation Quechua speakers are often represented as substituting high

vowels for mid vowels in their pronunciation of Spanish words (sin

uˇr for se˜nor ‘Sir’, etc.). In

reality, most speakers of Quechua have learned to master the distinction between high vowels

and mid vowels, except in word-ﬁnal position.

3.2 The Quechuan language family 197

*suqta [sɔqta] ‘six’, Jauja [suxta], Huancayo [suʔta ∼ su:ta]. In dialects that do not have

uvulars, the mid vowels are almost exclusively conﬁned to borrowed lexicon.

Long vowels (a:, i:, u:)are found in Quechua I and in Pacaraos Quechua. When

part of a lexical root, long vowels are most often the product of relatively transparent

sound changes if not of borrowing or onomatopoeic formations.

The role of vowel

length in the verbal conjugation and, to a lesser extent, in nominal possessive marking

is considerably more complex. It can have a clear morphemic value, as in the case of

the ﬁrst-person marker (e.g. Tarma Quechua wayi ‘house’, wayi-: ‘my house’), or it can

fulﬁll less tangible functions emerging at morpheme boundaries. For instance, in Tarma

Quechua ismikta:-ˇci- ‘to cause to stumble’ vowel length is triggered by the trisyllabic

structure of the root ismikta-; compare wata-ˇci- ‘to cause to tie’ from wata- ‘to tie’,

where this is not the case.

Long mid vowels (e:, o:) occur in most dialects that have vowel length. They often

occur in loans, where they reﬂect the stressed vowel of a Spanish model, e.g. Tarma

Quechua ke:da- ‘to stay’, ‘to become’ from Spanish queda ‘he/she stays’. In the Callej´on

de Huaylas (Ancash) long mid vowels often occur as reﬂexes of the diphthongs ay and

aw,asinwe:ˇco(:) from *wayˇcaw ‘a tyrant-ﬂycatcher (Agriornis montana)’. Some of the

Hu´anuco dialects exhibit optional lowering of long high vowels in word-ﬁnal position

(e.g. ni: ∼ ne: ‘I say’; miku: ∼ miko: ‘I eat’); cf. Sol´a (1967), Toliver (1987) and Weber

(1989).

In linguistic work related to Quechua the diphthongs aw, ay, uy, iw and iy are virtually

always analysed as consisting of a vocalic peak and a consonantal coda. Hence the

frequent observation in literature that the language has no vowel sequences. It is the

statement of a convention, rather than of a fact, because the theoretical argumentation

necessary to motivate this choice is seldom provided.

In virtually all modern literature on Quechua, sequences consisting of two full vowels

are analysed as if separated by a glide consonant, e.g. Quechua IIC suwa ‘thief’, tiyay

‘to sit’; QI San Pedro de Cajas (northern Jun´ın) rawu ‘snow’, wayi ‘house’. Similarly, as

in the case of the diphthongs, the possibility of analysing such sequences as containing

an automatic (non-phonemic) transition has not been duly explored.

Vowel sequences

do occur, however, in the Argentinian dialect of Santiago del Estero, which suffered the

loss of intervocalic *w and, therefore, contains sequences of like vowels. The vowels in

such sequences are articulated separately or as a long vowel with a descending tone, for

instance, in (*tawa >) taa [t´aʔa], [tˆa:] ‘four’ (Kirtchuk 1987).

Weber and Landerman (1985) have proposed an analysis of the Quechua long vowels as se-

quences of /V/ + /h/, an interpretation which can be defended from a strictly synchronic

perspective.

Howkins (1973) is one of the few examples of an interpretation in terms of vowel sequences.

An exception is found in one of Cerr´on-Palomino’s earliest works (Cerr´on-Palomino 1967).

198 3 The Inca Sphere

Vowel ﬂuctuation between i and a is widely attested in the present-day Quechua dia-

lects, most frequently in root-ﬁnal position. Examples are Pacaraos wawi, other Quechua

wawa ‘(woman’s) child’; Tarma pani, Cuzco pana ‘(man’s) sister’. In other cases, such as

warmi ‘woman’ alongside warma ‘child’, ‘youngster’ in Ayacucho and Cuzco Quechua,

the choice of the vowel is contrastive, and no ﬂuctuation occurs. Fluctuation between a

and u (Quechua I kanan against Cuzco Quechua kunan ‘now’) is less frequent.

Vowel loss occurs occasionally, e.g. in Cajamarca wanˇci- ‘to kill’ from *wan

u-ˇci-

‘to kill’, ‘to cause to die’. Several sorts of phonetic reduction, including vowel loss,

drastically transformed the phonological appearance of QIIB Chachapoyas Quechua

(Taylor 1975, 1994). It appears to be connected with a shift from penultimate to ini-

tial word stress, which has occurred in some of the other northern Peruvian dialects

as well.

Among the consonants, the series of stops and affricates, *p, *t, *ˇc, *ˇc

, *k, *q have

been affected by a number of shared or partially shared changes that can be characterised

as voicing, fricativisation and the introduction of a contrast between plain, glottalised

and aspirated stops (sometimes referred to as laryngealisation). In addition, the affricates

and the uvular stop *q suffered further modiﬁcations.

Voicing is widely found in the northern Quechua dialects, that is, in Ecuador and in

northern Peru (both in Quechua IIB, and in the Cajamarca and Ferre˜nafe dialects of the

QIIA subbranch). It affected, in particular, stops and affricates following a nasal, as in

the following forms from Cajamarca.

(1) *anˇca > and

ˇz

a ‘very’

(2) *punku > pungu ‘door’

The sixteenth-century lengua general of the Inca administration and the presumed

‘coastal’ dialect described by Santo Tom´as (1560a, b) knew voicing as well (cf. sec-

tion 3.2.2). Among the Quechua varieties assignable to the Quechua IIB subdivision,

only the language of the early seventeenth-century Huarochir´ı document shows no traces

of voicing.

Some central Peruvian Quechua I dialects (Tarma in the province of Tarma, Jun´ın;

Paccho in the province of Huaura, Lima) exhibit voicing of the grave stops *k and *p.

The conditions under which this change took place are most peculiar. As can be seen in

the following examples from Tarma, it affected prevocalic non-initial consonants unless

they were preceded by a nasal.

(3) *l

uˇcka- > luˇcga- ‘to slip’

(4) *uˇc

pa > uˇc

ba ‘ash’

(5) *apa- > aba- ‘to carry’

(6) *manka > manka ‘pan’

3.2 The Quechuan language family 199

In some Tarma Quechua sufﬁxes, stops became voiced after a nasal as well, e.g.

-guna ‘nominal plural’ (< *-kuna), as in wayi-n-guna ‘his/her/their houses’. The same

occurred with the initial consonant of the verb ‘to be’ ga-(< *ka-).

Geographically, the Tarma and Paccho dialects are far apart. Leaving aside their

behaviour in relation to voicing, they coincide with neighbouring varieties, not with

each other. The simultaneous occurrence of a sound change under such highly speciﬁc

conditions in two totally different places cannot easily be explained.

As we saw before, laryngealisation (glottalisation and aspiration) of stops and af-

fricates does not seem to be the product of regular sound change. These sounds are

generally considered exotic elements in Quechua, although they do occur as well in

Quechua roots not borrowed from an Aymaran language. Laryngealised stops and af-

fricates are found in a subgroup of the Quechua IIC dialect subdivision, which comprises

Cuzco Quechua and the Bolivian varieties. Their presence can have a sound-symbolic

meaning. As a rule, no more than a single laryngealised stop or affricate can occur in

a Quechua word. If it does, it must be the ﬁrst prevocalic stop or affricate in the root.

Quechua sufﬁxes do not normally contain laryngealised stops or affricates.

In the Ecuadorian central highlands, aspirated stops (not affricates) occur along with

plain stops. In the northern highlands (province of Imbabura) the reﬂexes of aspirated

*p and *k are fricative f and x, respectively:

(7) *paki- ‘to break’ > (Pichincha) p

aki-

> (Imbabura) faki-

(8) *qipa ‘afterwards’ > (Pichincha) k

ipa

> (Imbabura) xipa

Fricativisation (consonant lenition or weakening) is prominent in the dialects of the

Cuzco region and in Bolivia. This change has affected stops and affricates in syllable-

ﬁnal position. Its output varies both geographically and dialect-internally on the basis

of the phonological environment; examples (9) and (10) are from Cuzco Quechua.

(9) *aptay > hax

t’ay ‘to carry with a handle’

(10) *upyay > uxyay ∼ uxay ‘to drink’

Colonial sources, such as Gonz´alez Holgu´ın (1608), suggest that this has not always been the

case, e.g. tantta ‘bread’ (presumably [tant’a]), instead of modern t’anta.

An exception would be the progressive aspect sufﬁx -ˇc

a- in Grau, Apurimac (Torero 1964); cf.

Ayacucho -ˇcka-, Cuzco -sya-, -ˇsa-. In the subdialects of Puno and Arequipa, sufﬁxes borrowed

from Aymara preserve the glottalised and/or aspirated stops of the donor language, e.g. in tiy-

api-ˇci- ‘to cause to live together’, from Quechua tiya-ˇci- ‘to cause to reside’ and the Aymara

sufﬁx -t

api- ‘together’.

The palatal glide is optionally lost in Cuzco Quechua after a velar or a uvular fricative.

200 3 The Inca Sphere

Mannheim (1991: 55–6) observes a correlation between syllable-ﬁnal weakening, on

one hand, and the introduction of glottalisation and aspiration in syllable-initial positions

on the other. Nevertheless, Stark (1985b) reports that the northern Bolivian Quechua

variety of Apolo (province of Franz Tamayo, La Paz, Bolivia) was not affected by

syllable-ﬁnal fricativisation, although it has glottalised and aspirated stops. Landerman

(1994) makes the same observation for eastern Apurimac.

In Cuzco Quechua, syllable ﬁnal *p loses its labial character in sufﬁxes. Its reﬂex is

a uvular fricative, which can no longer be distinguished from the homophonous reﬂex

of *q:

(11) *muna-pti-n > muna-qti-n [munax

ŋ] ‘if he/she wants it’

(12) *runa-p > runa-q [runax

] ‘of a man’

The different orthographies currently in use for Cuzco Quechua and the Bolivian

dialects do not systematically reﬂect the newly formed fricative distinctions that obtain

in syllable-ﬁnal and intervocalic positions.

In Ecuadorian Quechua consonant lenition only affects the (merged) velar and uvular

stops:

(13) *pusaq > pusax ‘eight’

(14) *paˇc

ak > paˇcax ‘hundred’

The Proto-Quechua distinction between the affricates ˇc and ˇc

has been preserved

unmodiﬁed in the southern half and in the extreme north of the Quechua I territory

(all of Jun´ın, province of Pasco in Pasco, province of Yauyos in Lima, and province of

Sihuas in Ancash), in the Quechua IIA dialects of Ferre˜nafe, Cajamarca and Yauyos,

and in Quechua IIB Amazonas. Of the remaining dialects, several, such as Ayacucho

and Hu´anuco Quechua, have lost the distinction.

(15) *ˇc

aki > ˇcaki ‘foot’

*ˇcaki > ˇcaki ‘dry’

In Cuzco and Bolivian Quechua, the old distinction between the two affricates is

reﬂected by the presence of glottalisation in one of the members of the minimal pair just

exempliﬁed.

(16) *ˇc

aki > ˇcaki ‘foot’

*ˇcaki > ˇc’aki ‘dry’

However, there is no question of a regular change, as can be seen in example (17)

where *ˇc is reﬂected without glottalisation.

(17) *ˇcaka > ˇcaka ‘bridge’

3.2 The Quechuan language family 201

In Puno Quechua the old distinction between retroﬂex *ˇc

and alveopalatal *ˇc is partly

preserved in the fricative reﬂexes of these sounds that occur in syllable-ﬁnal position

(Cerr´on-Palomino 1986: 407–8).

(18) *aˇcka > aˇsk

a ‘many’

(19) *wiˇc

qay > wisq’ay ‘to close’

In part of the Quechua I dialects (and in Pacaraos), there has been a tendency to

enlarge the difference between the two affricates (which, admittedly, is not always

easy to perceive). In much of the northwestern part of the Quechua I area (Ancash,

northern and western Hu´anuco, provinces of Cajatambo, Huaral, Huaura and Oy´on

in Lima, province of Daniel Carri´on in Pasco), the alveopalatal affricate became an

alveolar affricate (c)orfricative (s), e.g. Callej´on de Huaylas caki,Pacaraos saki

‘dry’.

Subsequently, in most of Ancash, western Hu´anuco and in Cajatambo the retroﬂex af-

fricate was advanced to the articulatory position of the former alveopalatal, e.g. Callej´on

de Huaylas ˇcaki ‘foot’. In dialects where the affricates did not undergo the full sequence

of fronting shifts, the resulting gap was ﬁlled up by the introduction of an alveopalatal

affricate (ˇc) from other sources. As a result, the dialects of Yanahuanca (Daniel Carri´on,

Pasco) and Picoy (Huaura, Lima) distinguish three affricate positions: ˇc

, ˇc and c (Escobar

1967, Creider 1967).

An opposite development can be observed in the province of Concepci´on in the

department of Jun´ın. In that subdialect of QI Huanca, the alveopalatal affricate ˇc has

become a retroﬂex (e.g. *ˇcaka > ˇc

aka ‘bridge’), thus creating the space for a further

change of *l

to ˇc (see below). For an extensive treatment of this phenomenon see

Cerr´on-Palomino (1973a, 1989a).

Of all consonants in the stop–affricate series, the uvular stop *q has proven the least

stable. It may be the case that Quechua expanded into areas where this sound had not

been in use previously (for a discussion of this issue see Cerr´on-Palomino 1990). QIIA

Cajamarca and QI Callej´on de Huaylas are among the few dialects that have preserved

the uvular stop unmodiﬁed. In the Quechua IIB dialects *q merged with *k into present-

day k [k]. In Cuzco Quechua and in the Bolivian and Argentinian dialects, it was affected

by syllable-ﬁnal fricativisation (see above). In Bolivia, the plain (non-glottalised, non-

aspirated) uvular became a fricative in other positions as well.

The reﬂex of *q is also a uvular fricative in Ayacucho Quechua, in Pacaraos Quechua

and in the Quechua I dialects of northern Jun´ın. In the area of Tarma and Jauja its uvular

character became less prominent, leading eventually to a merger with the glottal fricative

*h.Inthe northeastern sector of Quechua I (Hu´anuco), the modern realisation of *q is

usually a voiced stop (often written as g).

202 3 The Inca Sphere

The loss of intervocalic *q was reported for Pachitea Quechua in Hu´anuco by Toliver

(1987). However, as we have seen before, the most drastic transformation of the ancient

uvular stop can be observed in the Huanca dialects of Huancayo and Concepci´on in

southern Jun´ın. There, the uvular stop at ﬁrst became a glottal stop. In some local

varieties it then turned into vowel length (internally and when syllable-ﬁnal) or zero (in

most other environments). In the village of Chongos Bajo the effect of the glottal stop is

preserved after a nasal n by the fact that the latter is represented by its velar allophone

in the combination n

[ŋʔ ∼ ŋ]; in this case, the presence of the glottal stop itself is

redundant.

(20) *sinqa > sinʔa [si

ŋʔa ∼ siŋa] ‘nose’

In the town of Chupaca, the loss of the glottal stop was total in this case. It also

triggered the loss of the preceding nasal, the combination being replaced by a glide

(Cerr´on-Palomino 1989a: 72–3).

(21) *sinqa > siya ‘nose’

Several verbal sufﬁxes with an internal *q (particularly, -ˇsqa ‘past participle’,

-rqa- ‘past tense’, and -rqu- originally ‘outward movement’, now several other mean-

ings) show a tendency to lose the q element in a large number of dialects. One ﬁnds

either synchronic variation (Ayacucho -r(q)u-, -r(q)a-, Ecuador -r(k)a-), complete loss

(Tarma -ru-,-ra-, -ˇsa), or obligatory presence of q (Ancash, Pacaraos, Puno -rqu-, -rqa-,

-ˇsqa/-sqa). It may mean that the variation is traceable to the proto-language. A simi-

lar case with an internal *k is -yku- (originally ‘inward movement’, now several other

meanings), which alternates with -yu- (or even -y-)inmany dialects.

The two sibilants *ˇs and *s have remained distinct in all Quechua dialects except

in most of the Quechua IIC branch and in the Colombian Ingano dialect, where the

two sounds have merged into s. The realisation of *ˇs is alveopalatal [ˇs] throughout the

dialects maintaining the distinction, except in Quechua I Huanca where it is a retroﬂex

or apical [ˇs

]. Additionally, the Huanca dialects have [ˇs] before i and in a limited number

of proper names (e.g. ˇsanti, Spanish Santiago ‘St James’). Quechua IIC Santiago del

Estero is different from the other Quechua IIC dialects in that it retains the distinction

between an alveodental and an alveopalatal fricative after i before a consonant (cf. de

Reuse 1986), as in (22):

(22) a. *iˇskay > iˇskay ‘two’

b. *ismuy > ismuy ‘to rot’

The fate of the alveodental *s has been much more varied than that of its alveopalatal

counterpart. In Pacaraos and in most of the Quechua I dialects (e.g. Ancash, northern

3.2 The Quechuan language family 203

Jun´ın) *s became a glottal fricative (h)word-initially; in syllable-ﬁnal position *s was

retained everywhere.

(23) *sara > hara ‘maize’

Word-medially, after a nasal consonant, *s became h in northeastern Ancash, and zero

in Callej´on de Huaylas and Tarma/northern Jun´ın, but it was retained in several other

Quechua I dialects (e.g. in Huanca).

(24) *kumsay > kumhay, kunhay ‘to push’

kumay

The Quechua I dialect of Jauja (Jun´ın) retains s in lexical items where all other

Quechua dialects have h or a reﬂex of *h.

(25) sana ‘upper part’

(compare Cuzco hanaq)

(26) sampiy ‘to cure’

(compare Cuzco hampiy)

The unique cases of s-retention in Jauja have been attributed to hypercorrection (Parker

1971: 59–60; cf. Cerr´on-Palomino 1989a: 28). It should be observed, however, that

several lexical items with initial h, such as hatun ‘big’, are never found with s instead

of h, not even in Jauja.

The changes affecting *s allow for several exceptions. Initial s was preserved in

a number of words with negative connotations (siki ‘buttocks’, sakwa- ‘to have sex-

ual intercourse’, supi- ‘to break wind’, supay ‘devil’, ‘evil spirit’, suwa ‘thief’) and,

rather persistently, in a few additional lexical items not belonging to the taboo sphere

(sapi ‘root’, sinqa ‘nose’). This phenomenon may be related to a mechanism of sound-

symbolism still productive in Tarma (northern Jun´ın) Quechua. It involves the frica-

tives s, ˇs, and h in lexical sets characterised by the presence of a pejorative member

with s,ahypochoristic member with ˇs, and a neutral member with h (Adelaar 1977:

290–2).

Some lexical items are notorious for exhibiting irregular reﬂexes of the sibi-

lants. The numeral ‘one’ and the verb ‘to come’ appear as huk (or its reﬂex) and

hamu-, respectively, in most of the Quechua IIC dialects, which otherwise preserve initial

s.However, a sibilant appears in Ecuadorian Quechua (ˇsux, ˇsamu-) and in the Argentinian

Santiago del Estero dialect (sux,butamu-). In Quechua I, we ﬁnd huk and ˇsamu- for

these items. A similar confusion surrounds the words for ‘name’ (QI northern Jun´ın huti;

QI Huancayo, ˇs

uti;QIJauja, QIIC suti; QIIB ˇsuti); and ‘young lady’ (QI northern Jun´ın,

eastern Ancash hipaˇs;QICallej´on de Huaylas, QIIB San Mart´ın ˇsipaˇs; QIIC sipas).