Modelling the Quechua-Aymara relationship: Structural features, sociolinguistic scenarios, and possible archeological evidence Pieter Muysken Centre for Language Studies (CLS) Radboud Universiteit Nijmegen Abstract In this paper I will explore various sociolinguistic scenarios of language contact which may be potentially invoked to account for the complex relationship between Quechua and Aymara, and indicate what would be the possible archeological evidence needed to back them up, if present. The relation between Quechua and Aymara, or more appropriately between the Quechuan and the Aymaran languages, has long been an issue of often heated discussion. The evidence for a separate origin of Quechua and Aymara and intensive borrowing is stronger than that supporting a common origin. O ne language may be assumed to have been „modeled‟ on the other. I will argue on linguistic grounds that most likely Aymara provided the model for Quechua. The precise nature of the contact relationship remains to be established, however, and this paper will describe and evaluate eight, not necessarily all mutually exclusive, scenarios from the language contact studies literature that might be invoked to account for it. Thus, results from current work on language contact contact will be brought to bear on deep-time historical linguistics. Finally, I will speculate on what would constitute possible archeological evidence for these scenarios. 1.
Introduction
The relation between Quechua and Aymara, or more appropriately between the Quechuan and the Aymaran languages, has long been an issue of often heated discussion among Amerindianist scholars and historical linguists. The reason is that the languages have a great deal in common, but not in the way that, say, Latin, French and Spanish do. Compare first (1a-c) for „the father sees his daughter‟ in Latin and its daughter languages: (1)
a.
el padre ve the.M father see.3s.PR
a AC
su 3PO
b.
le père voit the.M father see.3s.PR
sa fille 3PO.F daughter
c.
pater father.NO
filiam suam daughter.AC daughter.AC 3PO.F.AC
hija daughter
vedit see.3s.PR
S PANISH F RENCH L ATIN
Two things strike the eye: first of all (as has been known for a long time) the words in the three languages are similar, suggesting a historical relationship: (2)
pater padre père
filiam hija fille
suam su sa
vedit ve voit
Notice that the words are not identical, although they are clearly related.
Second, there are some broad correspondences between the grammatical arrangements in the three languages, such as the use of a possessive pronoun, the subject at the beginning of the sentence, the marking of (3 rd singular) person and (present) tense on the verb, the gender distinction. At the same time, there are many differences in the specific grammatical patterns: in Latin the verb is often at the end of the sentence, case marking (nominative and accusative) plays an important role, while in Spanish and French the verb is in the middle and only in Spanish we find (limited) use of an accusative marker, a. There is gender in all three languages, but in Spanish it is not marked on the possessive pronoun. In French and Spanish the pronoun precedes the noun, while in Latin it follows. Finally, French and Spanish, but not Latin, have an article. These features of the relation between Latin and its two descendants – similarity between the words and broad but not exact correspondences in the grammar – are typical for genetically related languages. The relation between Quechua and Aymara is of a different nature, however. Consider the following example (Cerrón Palomino 2007: 222): (3)
a.
pi-man-taq qayna wata qullqi-ta who-to-EMP yester year money-AC
qu-rqa-n give-PA-3
Q UECHUA
b.
khiti-ru-sa maya mara qullq(i) who-to-EMP yester year money
chura-yä-na give-PA-3
A YMARA
„To whom did he give money last year?‟
Here there is a very different relationship between the languages: most words and word endings are not at all related, and one word is almost or completely identical, qullqi „money‟. Furthermore, there is an almost exact correspondence between the individual endings of the two languages, exact for the absence of an accusative marker in the Aymara sentence. Thus, the relation between the languages is an enigma. I will depart from the assumption that the evidence for a separate origin of Quechua and Aymara and intensive borrowing is stronger than that supporting a common origin (disregarding a very distant common origin as postulated by Greenberg 1987). To explain their strong structural similarity, I assume one language to h ave been „modeled‟ on the other, and I will argue on linguistic grounds that most likely Aymara provided the model for Quechua. Even if we make these assumptions, the precise nature of the contact relationship remains to be established, however, and this paper will describe and evaluate eight, not necessarily all mutually exclusive, scenarios from the language contact studies literature. In this paper I will explore various sociolinguistic scenarios of language contact which may be potentially invoked to account for the complex relationship between Quechua and Aymara, and indicate what would be the possible archeological evidence needed to back them up, if present. Thus, results from current work on language contact will be brought to bear on deeptime historical linguistics. Finally, I will speculate on what would constitute possible archeological evidence for these scenarios. 2.
The relationship
2.1
The earlier debate
The relation between Quechua and Aymara, or more appropriately between the Quechuan and the Aymaran languages, has long been an issue of often heated discussion. Orr and Longacre (1968) and, using more principled arguments, Campbell (1995) have argued for a common
origin. Following Adelaar with Muysken (2004: 34-36), Heggarty (2005), and McMahon et al. (2005), I will assume that the evidence for a separate origin of Quechua and Aymara and intensive borrowing is stronger than that supporting a common origin. The (quite striking) evidence for the genetic link includes: the phoneme systems of the two language families are similar enough to allow for the reconstruction of a common proto-system (Orr and Longacre 1968) disregarding the large number of later borrowings between branches of the two families in both directions, about 20% of the core vocabulary is shared as noted, the morpho-syntax of both languages shows an uncanny resemblance on the level of the structural and semantic organization of the grammar (Cerrón Palomino 1994) However, other factors militate against postulating a genetic unit: the phonotactic patterning and very specific vowel deletion rules characteristic of the Aymaran languages as a group are not found in Quechua while a bit over 20% of the words in the core vocabulary correspond between the two language families, the remaining 80% or so do not at all (Adelaar 1986) this is all the more surprising since the 20% words are very similar if not identical in both language groups there are a great many very specific structural and semantic correspondences, but these do not extend to the actual shapes of the grammatical morphemes, which are all different core parts of verbal inflection do not correspond: these are highly unpredictable in Aymara, but fairly regular in Quechua the semantic fields covered by a group of 150 specific culturally relevant lexical items do not overlap in the two language families (Heggarty 2005). Possibly, some of these differences are due to later developments in the Aymara family. Initial correspondences may have been larger (Rodolfo Cerrón Palomino, p.c.). Altogether, the scenario that best fits the data is that two possibly unrelated or only very distantly related languages coexisted in the same area, most likely central Peru, for a long time and profoundly influenced one another. One of the two probably was restructured in fundamental ways under the influence of the other language, and was remodeled on the basis of its morpho-syntactic patterns without taking over most of the actual grammatical morphemes associated with these patterns. Apart from the early contacts that have affected both families in their entirety, there has been intensive subsequent contact between individual branches of the Quechuan and Aymaran families. The most striking effects are found in southern Quechua varieties. The Quechua of Cuzco and Bolivia probably has adopted the aspirated and glottalized stops in word initial position from Aymara (Mannheim 1991). Quechua dialects in the area of Arequipa and Puno (Peru) have adopted several Aymara verbal suffixes, inclusive of the accompanying vowel reduction rules (Adelaar 1987), possibly evidence of massive shift from AQymara to Quechua, or of prolonged bilingualism. In many areas of southern Peru and Bolivia there have been processes of language shift in rural communities from Aymara to Quechua (mostly) or from Quechua to Aymara. 2.2
Aymara as the model?
It is more likely that Aymara provided the model on the basis of which Quechua was restructured than the reverse. The first set of arguments for this claim comes from complexity. Several aspects of Aymara show greater complexity than Quechua, if we consider the protolanguages (Cerrón Palomino 2008; Adelaar with Muysken 2004): Aymara has complex morpho-phonemically triggered vowel deletion rules, which Quechua lacks. (Admittedly, there are complex vowel lowering rules in Quechua.) Aymara has a more complex phoneme system than Quechua, including aspirated and glottalized series (which in Quechua are only present in Aymara-influenced varieties, and only in word-initial position). Cerrón-Palomino (2008) gives more verbal derivational suffixes for Aymara (18), in his comparative survey, than for Quechua (14). These particularly concern the system of spatial suffixes. In Quechua the relation between subject marking, object marking and tense marking is fairly regular and transparent, while in Aymara the paradigms are highly irregular. Evidentiality in Quechua is marked through a series of regular clausal enclitics, while in Aymara its expression is in part through verbal derivational morphology, in part through enclitics. On other aspects, including phonotactics, stress, nominal morphology, and clausal enclitics, the differences in complexity between the languages are minor, or Quechua may be slightly more complex. Nonetheless, it would difficult, I think, to argue for an overall greater complexity for Quechua. A second set of arguments derived from the more original looking personal pronouns and verbal and nominal person inflection in Aymara, as compared to Quechua. Consider Table 1 (cf. Cerrón Palomino 2000: 193, 204 for the Proto-Aymara forms):
1 first singular 2 second singular 3 third singular 4 first plural inclusive Table 1:
Ay Ay nominal pronouns suffixes ŋa naya
Ay verbal suffixes ta
Qu pronouns ñuqa
Qu nominal Qu verbal suffixes suffixes y/-: ni
huma
ma
ta
qam
Yki
nki
up a
pa
i
pay
N
n
hiwasa
sa
tana
ñuqanchik
Nchik
nchik
Personal pronouns and verbal and nominal person inflection in Aymara and Quechua.
Several things can be noticed when we consider this array. First of all, the Aymara forms look more original, in particular the nominal suffixes. Second the Aymara verbal suffixes appear to be unrelated to the nominal suffixes, forming a separate paradigm. The Aymara pronouns are related to the nominal suffixes: (4)
1 2 3 4
naya hu-ma u-pha hiwa-sa
In Quechua there is no separate 4 th person pronoun rather the pronoun consists of the first person singular pronoun ñuqa and the [+2] plural marker – chik . The Quechua nominal and verbal suffixes resemble each other. Possibly the verbal system was derived through the combination of – n- and the nominal suffixes. 2.3
Evaluating the complexity argument
Can we reliably assume that the complex model provided the model for the simpler one (independently of the issue of whether Aymara is the more complex language)? Complexity is a complex issue in linguistics, since it has been a commonplace assumption in 20 th century linguistics that all languages are equally complex. Thus Bisang (2004) argues for the notion of „hidden complexity‟ i n those cases where the grammar of one language makes less overt distinctions (e.g. gender or number) than the other. However, there is also a tradition, going back to McWhorter (2001) and in effect Bickerton (1980), and including Dahl (2004), of defining complexity in terms of overt grammatical categories being expressed or not. Following this tradition, and using the WALS data base (Haspelmath et al. 2005), Parkvall (2008) argues that Creole languages are considerably less complex than other languages, on the whole. On the whole then, we can assume that intensive language contact may lead to less complex systems, but this is not necessarily the case for all types of language contact. Sometimes, a mixed language may combine or duplicate properties of two contributing languages. An example would be the Media Lengua from Salcedo, Ecuador, where the verb „show‟ can have three forms:
(5)
a. b. c.
mustramustra-chibi-chi-
(cf. Sp mostrar „show‟) (cf. mostra-chi- „show-CAU-„) (cf. Sp ver-chi- „see-CAU-„)
The form in (5a) is directly modeled on Spanish, the form in (5c) is a calque of the Quechua verb riku-chi -[see-CAU-] „cause to see‟, while the form in (5b) combines the Spanish lexical shape with the Quechua causative suffix. 2.4
What was pre-Quechua like?
A very difficult question concerns the nature of the language that could have been remodeled to yield present-day Quechua varieties. Two directions of investigation may point us to a putative answer. First, we can apply a simplistic formula of the type: Quechua – Aymara = pre-Quechua Put differently, if we take away the features in Quechua shared with Aymara, what do we have left? Second, once this has been done, we can look at internal evidence in the language, to see what could a plausible description of pre-Quechua. Obviously, any answers arrived at with these two strategies can only be extremely tentative. A first original Quechua feature may be a greater tolerance for closed syllables in roots and affixes (Adelaar 1986). Compare the following roots and affixes in both languages: Roots Q
Ay
Affixes Q
Ay
qam pay ñan huk iskay qanchis pusaq isqun pachak kay chay anqas
juma jupa
you s/he road maya one paya two paqallqu seven kimsaqallqu eight lla(:)tunka nine pataka hundred aka this uka that larama blue
-man -paq -wan -nchik -n -raq -pas -taq -mi/-m -si/-s
-ru -taki -mpi -täna -i -raki/-ra -sa -raki -wa -mna
Dative/illative Benefactive Instrumental/comitative 1 st plural inclusive 3r singular Continuative Additive Contrastive 1 st hand knowledge 2 n had knowledge
Table 2: Comparison of closed syllable roots in roots and affixes in both languages. A second feature comes from the object marking paradigm, which shows an inverse rd nd marker for the „3 subject > 2 o bject‟ relationship: (6)
riku-su-nki see-INV-2 „s/he sees you‟
It is not to be excluded that this reflects a trace of an earlier inverse marking system in the language. Willem Adelaar (p.c.) suggests that inverse marking may have been adopted by Quechua as a strategy to make the language more similar to the grammatically more complex neighbouring language Aymara, A third feature becomes apparent when we consider a verb like hamu- „come‟, which contains the associated motion particle – mu-. The element * ha- does not occur as a verb root in the language, although we do have suppletive exhortative forms such as haku „let‟s go‟. Other examples include miku- „eat‟ and michi- „let animals graze‟ (notice that – chi- is the causative suffix, and – ku- a medial suffix), while * mi- does not exist as a root. It is not impossible that there was a tendency towards mono-syllabic roots in pre-Quechua, and that many roots were transformed into bi-syllabic words under the influence of Aymara. Willem Adelaar (p.c.) also mentions * qa- „float‟; * ya- „go‟ or „move‟; *wa- „hang‟; *tru- „place‟, „put‟; *(s)ha- „stand‟, „move vertically‟; possibly * ta- „sit‟ or „plant‟. Currently, there is one monosyllabic verb in Aymara, sa- „say‟(saha- in the related language Jaqaru), while Quechua has a number of mono-syllabic verbs: (7)
niqurika-
„say‟ „give‟ „go‟ „be‟
If the analysis given above is correct, we may add ha- „move‟ and mi- „eat‟ to this list, and a further analysis of the Quechua lexicon could reveal more potential mono-syllabic roots. Following this line of reasoning, we thus have three potential Quechua features predating its contact with Aymara: Inverse marking Possibility of closed syllables in roots and affixes
Possibility of monosyllabic roots On the whole, however, pre-Quechua may have been similar to Aymara in its word order, for instance; there we do not find any major differences nor evidence of structuring. It is also likely, if the contact scenario that we are adopting is correct, that preQuechua may have lost features (e.g. person prefixes) when coming into contact with Aymara. I do not know, however, how this loss could be detected, given the overall harmonious typological profile of Quechua (see section 3.5 below). Willem Adelaar (p.c.) suggests that the – m in qam „you‟ may currespond to a 2 nd person prefix, and that the ñu- in st ñuqa „I‟ may have its roots in a 1 person prefix as well. 3.
Contact scenarios
3.1
Overview
In the language contact literature, a number of scenarios have been discussed, summarized in Table 3. In this table, I have divided them up along the lines of the frequently made distinction between shift, maintenance, and symmetrical scenarios (cf. Thomason and Kaufman 1988). Label
Key references
Examples of language settings
Definition
Muysken 1981, 1996a Keesing 1988 Lefebvre 1998 Schwartz and Sprouse 1996; van de Craats et al. 2000, 2002 Von Wartburg 1939
Media Lengua
Roots replaced, by elements from a new language, affixes retained. All elements with a new shape from the dominant language Patterns from the first language are retained as people are acquiring a second language
Shift scenarios
Relexification 1 Relexification 2 L1 transfer
Substrate
Melanesian Pidgin; Haitian Creole Immigrant settings in Western Europe Celtic and Basque in Vulgar Latin, yielding French and Spanish
As populations shift to a new conquering language, patterns of their original languages influence the outcome
Global phenomenon
Words come in from another language, often with prestige, and with these words affixes and idioms are taken over A small group retains its language, but also speaks a dominant neighbouring language, from which patterns are then taken Roots replaced,by elements from a traditional language, affixes retained Languages adopt words, phrases, and idioms from an (often international) prestige language
Maintenance scenarios
Borrowing Metatypy or restructuring
Poplack, Sankoff, and Miller 1988 Ross 1999; Winford 2003
Relexification 3
Muysken 1996b
Superstrate
Von Wartburg 1939
Australia; New Guinea; Turkey; etc. Kallawaya Latin in Western Europe; French in medieval England
Symmetrical scenarios
Sprachbund convergence
Table 3:
Emeneau 1956
Balkan; India
Languages influence each other mutually, and start adopting common structures
Systematic overview of different contact scenarios
I will discuss these clusters of scenarios in terms of grouping given in Table 3.
3.2 Shift scenarios In shift scenarios the general idea would be that there is an original population of speakers that adopts a second language but in doing so, transforms it, so that new language structures emerge. A number of distinct but related scenarios go under names like relexification, L1 transfer, and substrate influence. I will begin with various kinds of relexification. The first type of relexification, which I will label here Relexification 1 (Muysken 1981, 1997a), is characteristic of several regions of Ecuador. In the local Quechua, all roots have been replaced with Spanish phonetic shapes, while the affixes have been retained and the semantic and syntactic properties of the original root lexemes remain largely intact. The resulting variety is often called Media Lengua, although locally several other names are used. An example would be (8) (cited from Muysken 1997a: 388), where the first line is Media Lengua, the second line the corresponding form in the local Quechua variety, and at the bottom the Spanish equivalent is given: (8)
nustru-ga alla-wa-bi-mi ñukunchi-ga chay-wa-bi-mi 1pl-TO that/there-DIM-LO-AF
sinta-nchi tiya-nchi sit-1pl
„We live over there.‟ „Nosotros vivimos allá.‟
MEDIA LENGUA Q UECHUA SPANISH
Notice that the Media Lengua example follows the Quechua model exactly. The same word order is the same and the Quechua suffixes reappear in Media Lengua. The Spanish verb used for „live‟ is not Spanish vivir but rather sentar „sit‟, just like the Quechua verb tiya-. The process of Relexification 1, besides Quechua in Ecuador, has been documented for Griekwa Afrikaans. The community language is retained, but the lexical shapes froma dominant language are brought in. It has been associated with the creation of mixed identities, acculturation, and language shift. This type of relexification have the plus feature that it entails a close structural similarity between copy and model, which we need to explain the Quechua/Aymara relationship. However, it fails in that the non-root elements (affixes and clitics) of copy and model will result to be identical in their actual shape, which is decidedly not the case for the language pair I am considering in this paper. The result is better for what I will here label Relexification 2 (Keesing 1988; Lefebvre 1998), a process postulated to account for the genesis of pidgins and creoles, following an old idea that these are the result of a match between European colonial lexicons and indigenous (African, Pacific) syntax and semantics. An example involving Fongbe and Haitian Creole from Lefebvre‟s work (1976: 282) is : (9)
Mari te prepare Mari kò Dà Mary ANT prepare
pat wÒ dough
HAITIAN CREOLE FONGBE
„Mary prepared dough.‟
In this example, Lefebvre argues that the Haitian Creole anterior tense form te is modeled on Fongbe Dà, even though etymologically derived from French était . Similarly Keesing (1988) claims that many structures in Melanesian Pidgin derive directly from Oceanic models. An Solomon Islands Pidgin example, with a Kwaio model, would be (10): (10)
mi
no
luk-im iutufala
PIDGIN
ku „ame aga-si-„amo‟o I NEG see-TR you2 „I did not see the two of you. ‟
KWAIO
Many parallel examples can be adduced from the Pacific in Keesing‟s work. While Lefebvre‟s claim that structural and semantic properties of Haitian Creole can be
systematically attributed to Fongbe patterns has been criticized, it is clear that Relexification 2 processes have played a role in the genesis of the Atlantic and Pacific creoles. On the plus side, Relexification 2 clearly satisfies the two requirements posed by the Quechua-Aymara relationship: close structural similarity and morphemic distinctness. On the minus side, the cases documented all involve isolated words, both lexical and grammatical, however, and not the complex agglutinative patterns shown by the language pair I am focusing on. Very similar to Relexification 2, and for some researchers identical to it, is the process of Full Transfer of features of the first language (L1) in learning a second language (L2), also referred to as Conservation (Schwartz and Sprouse 1996; van de Craats et al. 2000, 2002). This approach departs from the assumption, derived from the Principles and Parameters framework (Chomsky 1981), that children acquiring their first language fix a number of structural features, labeled parameters. These adopted parameters are then fully transferred to a second language in the L2 acquisition process, until the learner gradually adopts the structural features or parameters of the second language. Thus the initial stages of second language learning reflect a structural replica of the first language, in this view. An example would be the Turkish possessive construction in (11a), its initial reflections in different stages of Dutch interlanguage in (11b), and the Dutch model form in (11c) (Van de Craats et al. 2000):. (11)
a.
Ayse-nin Ayse-GEN
araba-si car-3
TURKISH
„Ayse‟s car‟
b.
i. Ayse auto ii. Ayse van auto iii. van Ayse auto iv. auto van Ayse
c.
de auto van Ayse the car of Ayse
[Ayse car] [Ayse of car] [of Ayse car] [car of Ayse]
DIFFERENT STAGES OF T URKISH-DUTCH INTERLANGUAGE
DUTCH MODEL
In this sense, full transfer is very similar to Relexification 3. It is different in that there is no claim in this hypothesis that the lexical semantic features of the model language words are retained in the process of acquisition. It is a hypothesis about purely structural properties. This is actually a bonus for this model with respect to the Quechua-Aymara relationship, since Heggarty (2005) has shown that there is no strict lexical semantic correspondence between Aymara and Quechua content words. On the minus side, the Full Transfer/Conservation Hypothesis has not been applied so far to cases where agglutinative suffixing morphology is retained from the first language. Notice in this respect that in (11b) we get no morphological reflection of the possessive suffix on the noun. Substrate is a notion from historical linguistics. It has been used to explain why the off-shoots from Latin, e.g. French and Spanish, differ so much from each other, for instance in their pronunciation. For French it is claimed that a Celtic language speaking population
transferred some of their pronunciation features to their second language Vulgar Latin, while for Spanish Basque speakers are claimed to be involved. The original term comes from Von Wartburg (1939), but the notion of substrate is now widely accepted as an important process in historical language development. Many examples of substrate influence have been brought forward in different language settings. In the Andes, there is clear evidence for Quechua and Aymara substrate influence in Andean Spanish (Adelaar with Muysken 2004). Its relation to second language acquisition is also clear; however, most postulated cases of substrate influence do not involve thorough restructuring of the model language. If we consider for a moment the social and demographic factors which may lead to (a) considerable restructuring of the model language, and (b) preservation of features of the source language, the following come to mind: (a) Homogeneous substrate. From the literature on pidgins and creoles it is clear that the more similar the speech varieties of the original population (either they all speak the same language or structurally very similar languages), the more likely that features of their language persist in the process of shift. (b) Limited access to model or target . A factor which is often invoked to explain heavy restructuring of a model is the amount of access to the model, in terms of absolute numbers of speakers of the model and of actual day to day interaction with those speakers. The more limited the access to the model, the more restructuring will ensue. (c) Covert prestige of original language and creation of separate identity . For the linguistic variety resulting from language shift to be clearly distinct from the model, it is important that the original group of speakers retain a strong sense of their original identity, or create a new identity, distinct from that of the model group. When we apply these considerations to the Aymara-Quechua relationship under the shift perspective, we have: a group of Aymara speakers brought into contact with a dominant group of pre-Quechua speakers, with whom they have only limited direct contact adopting a form of Quechua as their new language without however blending into the culture of the larger pre-Quechua group Given the degree to which Quechua resembles its supposed Aymara model, we would have to assume that all three factors (a)-(c) were satisfied under the shift scenario, to produce the effect needed. 3.3
Maintenance scenarios
As the name already indicates, in the maintenance scenarios the original language of the community is maintained, at least as an emblem of ethnolinguistic identity, but then undergoes changes due to surrounding languages. The best known process in this cluster is of course lexical borrowing , a process extremely well researched (the most detailed study is perhaps Poplack et al. 1988). It is found in all languages, as far as known. While incidental cultural borrowing is exceedingly common, the more intensive borrowing, affecting a variety of word classes as well as core lexical items, is less frequent. It is certainly a central process in subsequent Quechua-Aymara contact history (cf. e.g. Adelaar 1987), but cannot help us explain the complex interrelations which are the subject of this paper. Even if we invoke the very heavy borrowing of Italian and Sicilian into Maltese Arabic (Aquilina 1959) or of Spanish into Chamorro (Stolz 2003), we end up with the wrong results. Words may be borrowed, some affixes, possibly idiomatic
semantic patterns, but not really structural features, and certainly not on the scale needed to explain the Quechua-Aymara similarities. A more promising possibility is Metatypy (Ross 1999) or contact-induced Restructuring (Winford 2003). Following Ross (1999: 7), Metatypy is a „change in morphosyntactic type and grammatical organisation (and also semantic patterns) which a language undergoes as a result of its speakers‟ bilingualism in another language.‟ Ross has carried out research on Karkar island in the Pacific and notices that the western Oceanic language Takia has undergone significant structural influences from the neighbouring Papuan Waskia language, presumably from the Trans-New Guinea Phylum. The result is that words in the two languages can be matched can now be matched one by one: (12)
tamol man kadi man
an DET mu DET
ŋai
me aga me
i-fun-ag=da he-hit-me=IMP umo-so hit-PR.he
TAKIA WASKIA
„The man is hitting me.‟
Takia has taken over verb final and [noun determiner] word order from Waskia, and thus hasd changed its typological profile. Notice, however, that the affix order has not been affected. In Takia person markers precede the verb, while in Waskia the subject marker follows the verb. The terminology used by Malcolm Ross is that of Modified language (the weaker language, Takia) and Inter-community language (the stronger language, Waskia). Other putative cases of metatypy (among many others) are listed in Table 4: Modified language
Inter-community language
Dravidian Kannada Indo-Aryan Marathi Asia Minor Greek Turkish Mixe Basque Gascon Table 4:
Reference
Gumperz and Wilson 1971 Dawkins 1916 Haase 1992
Representation of some putative cases of metatypy
An Andean case would be, as noted by Cerrón Palomino (p.c.), the Uchumataqu (Uru) of Irohito (Bolivia), which in the last half a century of its existence was profoundly influenced structurally by the stronger neighbouring language Aymara (cf. e.g. Muysken 2000). Altogether the Metatypy scenario can shed light on the Aymara-Quechua relationship, but even in the paradigm case of Takia the morphology of the Modified language is not affected by metatypy. It mostly concerns the word order. However, the case of Asia Minor Greek reported on by Dawking (1916), cited in Thomason and Kaufman (1988), does indicate restructuring of the morphological system as well. The original Greek had inflectional morphology, but in several Capadocian villages forms of Greek developed with agglutinative morphological endings. Compare the following nouns from the Ferték variety (Dawkins 1916: 114, cited from Thomason and Kaufman 1988: 219): Turkish qız „girl‟ Sg. Nom. qız girl girl‟s Gen. qız-ın Pl. Nom. qız-lar girls Gen. qız-lar-ı (sic) girls‟
Ferték néka „woman, wife‟ néka wife wife‟s néka-yu néka-es wives wives‟ néka-ez-yu
Table 5:
Nominal paradigms in the Cappadocian Greek dialect of Ferték and possible Turkish models
Notice that here it is the agglutinative morphological pattern that has been taken over, not the actual Turkish affixes. Elsewhere, some cases of actual Turkish affixes that have been taken over are cited. A further possibility is what I will label Relexification 3 here. It closely resembles the earlier described Relexification 1 in structural terms. It can be illustrated with Kallawaya (Muysken 1997b), a ritual curing language from northern Bolivia. In Kallawaya typically Quechua morphology and syntax is used, but the lexical roots are non-Quechua. As far as we can establish (but little is known of this now extinct language), they are often derived from Puquina. A Kallawaya example is (13) (cited from Oblitas Poblete 1968: 40), with the Quechua equivalent in the second line: (13)
mii-qa runa-qa man-TO
llalli oja- ku-j-mi allin miku-(ku-)j-mi good eat-RE-AG-AF
acha-n ka-n be-3
KALLAWAYA QUECHUA
„The man is a very greedy eater.‟
The process of Relexification 3 is structurally very similar to Relexification 1, but differs from it in that it is not a new or dominant language that provides the lexical roots, but rather the ancestral community language. It has also been documented for the Para-Romani language varieties in Europe, such as Anglo-Romani and Caló (Bakker and Van der Voort 1991), where Roma groups adopted words from their original language as part of the new languages of the countries where they had settled. Relexification 3 is not useful for the purposes of this article because concrete morphological material from the new language is adopted, which cannot have been the case if Quechua vocabulary was retained in a relexification with Aymara structures. The converse notion of substrate influence is Superstrate influence, exerted by a dominant language on one or more dominated languages. Examples abound. French was dominant over Anglo-Saxon varieties in medieval England, Latin was a superstrate language in Renaissance Western Europe. French as a superstrate in 18 th century and early 19 th century Europe. The influence of a superstrate language is primarily lexical, and in some cases it involves calqued idioms as well (cf. French s´il vous plait > English if you please , Dutch als ‘t u belieft ). In the Andes I can mention both Spanish superstrate influence on Quechua and Quechua influence on languages such as Uru, Shuar, and Cholon. Superstrate influence is a very common scenario, but it is unlikely that it played a significant role in the remodeling of Quechua. Thus of the various maintenance scenarios, the only serious candidate is the metatypy of restructuring scenario. Relevant social factors for it are: (a) There is a small community of speakers of the Modified language, surrounded by or in close contact with speakers of the much larger and stronger Intercommunity language; (b) Speakers of the Modified language form a sufficiently tight network to be aware of their separate identity; (c) Their language is a marker of that identity; (d) Many speakers of the Modified language are bilingual in the Inter-community language and use it so often that they come to know it better than the emblematic original language of the community. When we apply these considerations to the Aymara-Quechua relationship under the maintenance perspective, we have:
a small group of pre-Quechua speakers brought into contact with a dominant group of Aymara speakers, with whom they are in frequent and direct contact maintaining a form of Quechua as their community language however taking over the patterns and much of the culture of the larger Aymara group Given the degree to which Quechua resembles its supposed Aymara model, we would have to assume that all three factors (a)-(d) were satisfied under the maintenance scenario, to produce the effect needed. 3.4
Symmetrical scenarios for language contact
Another issue altogether is whether we need to assume asymmetry in the contact relationship, as was the case under the maintenance and the shift scenarios. Maybe Pre-Proto-Quechua and Pre-Proto-Aymara (“Pre-Proto” simply refers here to the situation of the proto -languages before contact) were simply neighbours, possibly practicing linguistic exogamy (Sorensen 1967; Jackson 1983; Aikhenvald 1999, 2002; Epps 2007), and influenced each other heavily in bilingual communities or families. Under a symmetrical scenario there is mutual influence rather than unidirectional modeling. Of course it is difficult to exclude this possibility altogether, but there are two structural properties of both Quechua and Aymara that speak against it. First of all, typological consistency is high in both languages. Typological consistency (cf. e.g. Lehmann 1973) is a somewhat controversial notion. It assumes that a certain configuration of structural features represents something of an „ideal‟ type in languages. A historical concomitant of the
assumption would be that languages that do not conform to the ideal type are somehow underway between two ideal types, and will eventually become ideal again in their structural make up. This historical hypothesis has been rightly criticized because it cannot be falsified. Also, it is not clear why languages would sometimes depart from such an ideal state. Nonetheless, many if not most linguists would agree that both Quechua and Aymara are typologically consistent. They share the following properties: (14)
Verb final order Postpositions and case suffixes Pre-nominal adjectives and determiners Pre-nominal possessors Suffixing morphology Head final compounds
They share these features with typologically equally „consistent‟ languages like Turkish and
Japanese, which has led some researchers to propose a genetic relationship with these languages, at various points in time. This consistency would be hard to explain if the structural similarity of the two languages were the result of symmetrical blending of thes languages. In such a case a typologically inconsistent pattern would be much more likely. A second, possibly related, argument concerns the parallelism between the nominal and verbal domains in the two languages. In nominalizations, the parallels actually lead to a merger of nominal and verbal properties (for Quechua, cf. Lefebvre and Muysken 1988). In mixed language resulting from the blending of two distinct systems, the two systems are very likely to be much more distinct. Examples include the language of the Canadian Plains Michif (Bakker 1997), in which Cree verb phrases are combined with French noun phrases, each
keeping its own properties. In Australia, several mixed languages have emerged including Light Warlpiri (O'Shannessy 2005) and Gurindje Kriol (McConvell and Meakins 2005), where case marking systems from aboriginal languages are combined with English verbal and auxiliary elements, in a sentence frame partially based on aboriginal models. In Copper Island Aleut (Golovko and Vakhtin 1990), finally, Russian pronouns and verb inflections are marked with Aleut grammar and nominal morphology. In all these cases, the system in the nominal constituents is very different from the one in the verbal constituents typologically. These two arguments are not conclusive. With respect to the first argument, the notion of typological consistency may be a relative notion, the secondary effect of processing constraints (Hawkins 1994), and thus either or both Quechua and Aymara may have drifted towards this ideal order pattern independently, even after initial contact. Also, if typological pressure is strong enough to create strong enough resemblances between such languages as Quechua and Turkish for researchers to claim a common origin for both (e.g. Hakkola 1997), why could this not also have been the case for Quechua and Aymara? The answer would be here that Aymara and Quechua share many similarities much beyond the typological profile listed in (8), which could not be explained through typological pressure. At the very least there is no linguistic evidence which suggests that the structural similarities of the two languages are due to symmetrical mutual convergence. For this reason, I will no longer pursue this option in the following discussion. 3.5
Summary
We are thus left with two very different, in most ways radically opposed, perspectives. In the shift perspective, an Aymara-speaking group came into contact with a dominant pre-Quechua speaking group, acquired its language, but maintained the grammatical patterns of its own language. Perspective Original group Dominant new group Nature of contact between groups Direction of language change
Table 6:
Shift
Maintenance
Aymara Pre-Quechua Limited Towards Quechua shapes
Pre-Quechua Aymara Intense Towards Aymara patterns
Main features of the two perspectives on the possible Quechua remodeling on the basis of Aymara
On the basis of (my reading of) the available studies of language contact, a maintenance scenario, in which a smaller pre-Quechua speaking group came to be dominated a stronger Aymara-speaking group, is the more likely one, involving the process of metatypy. This remains open to discussion of course. For Willem Adelaar (p.c.), the shift scenario remains more attractive than the maintenance scenario. One could imagine that a part of the Pre- Aymara‟s shifted to PreQuechua, whereby the latter was extensively remodeled. The rest kept speaking (Pre-) Aymara. The not remodeled part of Pre-Quechua presumably would have died out. It should be stressed that this only refers to the initial contact between the two languages. It is clear that later on there have been multiple cases of maintenance, shift, and prolonged coexistence of the two languages in various locations. 4.
Possible archeological evidence
If we want to bring archeological evidence to bear on the question of how to choose between these scenarios, or more simply, to connect historical linguistic considerations to archeological data, we would need to be able to: link Aymara-speaking and pre-Quechua speaking cultures to specific archeological remains. In addition, the archeological sites where elements from both cultural traditions are found ideally should be able to inform us about: proportions of speakers in particular stages of contact degree of integration that accompanied the contact at those stages Returning for a moment to the two dominant types of scenarios, Table 7 gives an overview of the relevant archeological evidence needed to support either scenario. Perspective Original group
Shift
Maintenance
Remains of Aymara materials
Dominant group
Some evidence of Pre-Quechua materials, which are strongly present elsewhere and associated with a strong cultural tradition Limited presence of those materials in the originally Aymara community
Evidence of a Pre-Quechua cultural tradition in a specific locality Progressive increase of Aymara cultural elements
Nature of contact between groups Direction of change
Table 7:
Progressive increase of Quechualinked materials, but evidence of a transformation dynamic
Replacement at the site of preQuechua with Aymaran materials A local form of the Aymara cultural tradition
Main features of possible archeological evidence for the two perspectives on the possible remodeling o Quechua
If indeed a maintenance scenario is the most appropriate one, the third column is what we should be looking for. 5.
Concluding remarks
In this paper I have tried to clarify the various possibilities to explain the relation in terms of language contact between Quechua and Aymara, both in terms of a first examination of their linguistic characteristics, and in terms of the literature on language contact. I have also tried to suggest possible consequences for what we should find in the archeological record. The results point in a certain direction, metatypy, but this remains very much open to question, as the precise nature of the relation between the two languages is further elucidated and the possible relevance of other contact studies is further explored. Note
I am grateful to Willem Adelaar and Rodolfo Cerrón Palomino for detailed comments on an earlier draft. Needless to say, any error of interpretation or representation of their materials remains my responsibility. References
Adelaar, Willem F. H. (1986) La relación quechua-aru: perspectivas para la separación del léxico. Revista Andina 4, 379-426. Adelaar, Willem F. H. (1987) Aymarismos en el quechua de Puno. Indiana 11, 223-231. Adelaar, Willem F.H. with Pieter C. Muysken (2004) The languages of the Andes . Cambri1dge: Cambridge University Press. Aikhenvald, Alexandra Y. (1999) Areal diffusion and language contact in the Içana-Vaupés basin. In Dixon and Aikhenvald (eds.), 385-416. Aikhenvald, Alexandra Y. (2002) Language contact in Amazonia. Oxford: Oxford University Press. Aquilina, Joseph (1959) The structure of Maltese: A study in mixed grammar and vocabulary . Valletta (Malta): The Royal University of Malta. Bakker, Peter (1997) A language of our own. The genesis of Michif, the mixed Cree-French language of the Canadian Métis . Oxford University Press. Bakker, P. & Van der Voort, H. (1991) Para-Romani languages: An overview and some speculations on their genesis. In: In the Margin of Romani. Gypsy languages in Contact , edited by Bakker, P. & Cortiade, M. Amsterdam: Publikaties van het Instituut voor Algemene Taalwetenschap nr. 58. (Studies in Language Contact 1), pp. 16-44 Bickerton, Derek (1980) Roots of language. Ann Arbor, Mich.: Karoma. Bisang, Walter (2004) “Grammaticalization without Coevolution of Form and Meaning: The Case of Tense-Aspect- Modality in East and Mainland Southeast Asia.” In Walter Bisang, Nikolaus Himmelmann and Björn Wiemer (eds.), What Makes Grammaticalization? — A Look from its Fringes and its Components . Berlin: Mouton de Gruyter, pp. 109-138.
Campbell, Lyle (1995) The Quechumaran hypothesis and lessons for distant genetic comparison. Diachronica 12, 157-200. Cerrón-Palomino, Rodolfo (2000) Lingüística aimara. Cuzco: Centro Bartolomé de las Casas. Cerrón-Palomino, Rodolfo (2007) Quechumara. Estructuras paralelas del quechua y del aimara . La Paz: Plural Editores. Chomsky, Noam (1981) Lectures on Government and Binding. Mouton de Gruyter.
Craats, Ineke van de, Norbert Corver, and Roeland van Hout (2000) Conservation of grammatical knowledge: On the acquisition of POSSessive noun phrases by Turkish and Moroccan Arabic learners of Dutch. Linguistics 38, 221-314. Craats, Ineke van de, Roeland van Hout, and Norbert Corver (2002) The acquisition of possessive HAVE-clauses by Turkish and Moroccan learners of Dutch. Bilingualism: Language and Cognition 5, 147-174. Dahl, Östen. 2004. The growth and maintenance of linguistic complexity. AmsterdamPhiladelphia: Benjamins. Dawkins, Richard M. (1916) Modern Greek in Asia Minor. A study of dialect of Silly, Cappadocia and Pharasa . Cambridge: Cambridge University Press. Emeneau, Murray. 1956. India as a Linguistic Area. Language 32, 3-16.
Epps, Patience (2007) The Vaupés melting pot: Tukanoan influence on Hup. In Aikhenvald and Dixon (eds.), 267-289. Golovko, Evgeniy V., and Nicolai B. Vakhtin (1990) Aleut in contact: the CIA enigma. Acta Linguistica Hafniensia 22, 97-125. Greenberg, Joseph H. 1987 Language in the Americas. Stanford: Stanford University Press. Gumperz, John J. and Robert Wilson (1971) Convergence and creolization: a case from the Indo-Ayan / Dravidian border in India. In D. Hymes (ed.) Pidginization and creolization of languages . Cambridge: Cambridge University Press, 151-167. Haase, Martin (1992) Sprachkontakt und Sprachwandel im Baskenland: Einflüsse des Gaskognischen und Französischen auf das Baskische . Hamburg: Buske Hakola, Hannu Panu Aukusti (1997) Duraljan vocabulary: lexical similarities in the major agglutinative languages . Kupo, Finland: private edition. Haspelmath, Martin, Matthew S. Dryer, David Gil, and Bernard Comrie (eds.) World Atlas of Linguistic Structures . Oxford: Oxford University Press.
2005
Hawkins, John A. (1994) A performance theory of order and constituency . Cambridge: Cambridge University Press. Heggarty, Paul (2005) Enigmas en el origen de las lenguas andinas: aplicando nuevas técnicas a las incógnitas por resolver. Revista Andina 40, 9-80. Cuzco: Centro Bartolomé de las Casas. Jackson, Jean 1983 The Fish People: Linguistic exogamy and Tukanoan identity in Northwest Amazonia. Cambridge: Cambridge University Press. Keesing, Roger M. (1988) Melanesian Pidgin and the Oceanic Substrate . Pasadena, Cal.: Stanford University Press.
Lefebvre, Claire (1996) The tense, mood, and aspect system of Haitian Creole and the problem of transmission of grammar in creole genesis. Journal of Pidgin and Creole Languages 11, 231-311. Lefebvre, Claire (1998) Creole genesis and the acquisition of grammar. The case of Haitian Creole. Cambridge: Cambridge University Press. Lefebvre, Claire and Pieter Muysken (1988) Mixed Categories: Nominalizations in Quechua . Dordrecht: Kluwer Academic Publishers. Lehmann, Winfred P. 1973. A structural principle of language and its implications. Language 49, 47-66. McConvell, Patrick and Felicity Meakins. 2005, 'Gurindji Kriol: A mixed language emerges from code-switching', Australian Journal of Linguistics. 25: 9-30. McMahon, April, Paul Heggarty, Robert McMahon & Natalia Slaska (2005) Swadesh sublists and the benefits of borrowing: an Andean case study, in: McMahon, April (ed.): Quantitative Methods in Language Comparison. Transactions of the Philological Society 103, 147-169. Oxford: Blackwell. McWhorter, John H. (2001) The world‟s simplest grammars are creole grammars, Linguistic Typology 5, 125-166.
Muysken, Pieter (1981) Half-way between Quechua and Spanish: the case for relexification. In Arnold Highfield and Albert Valdman (eds.) Historicity and variation in creole studies. Ann Harbor: Karoma, pp. 52-78. Muysken, Pieter (1997a) Media Lengua. In Sarah G. Thomason (ed.) Contact languages. A wider perspective. Amsterdam: Benjamins, pp. 365-426. Muysken, Pieter (1997b) Callahuaya. In Thomason, Sarah G. (ed.): Contact languages: A wider perspective . Amsterdam: Benjamins, 427-447. Muysken, P. (2000) Drawn into the Aymara mold? Notes on Uru grammar. In H. Van der Voort and S. Van der Kerke (eds.) Essays on indigenous languages of Lowland South America. Leiden: CNWS, pp. 99-109. O'Shannessy, Carmel (2005) Light Warlpiri: A new language. Australian Journal of Linguistics 25: 31-57. Orr, Carolyn J. and Robert E. Longacre (1968) Proto Quechumaran. Language 44, 528-555. Oblitas Poblete, Enrique (1968) El idioma secreto de los incas (vocabulario castellanocallahuaya) . La Paz. Parkvall, Mikael (2008) The simplicity of creoles in a cross-linguistic perspective. In, Miestamo, Matti, Kaius Sinnemäki and Fred Karlsson (eds.), Language Complexity. Amsterdam: Benjamins, pp. 265 – 285.
Poplack, Shana, Sankoff, David, and Miller, Christopher (1988) The social correlates and linguistic processes of lexical borrowing and assimilation. Linguistics 26, 47-104. Ross, Malcolm D. (1999). Exploring metatypy: How does contact-induced typological change come about? (Keynote talk, Australian Linguistic Society's annual meeting, Perth).
Schwartz, Bonnie D. and Richard A. Sprouse (1996) L2 cognitive states and the full transfer, full access model. Second Language Research 12, 40-72. Sorensen, Arthur P. (1967) Multilingualism in the Northwest Amazon. American Anthropologist 69, 670-684. Stolz, Thomas (2003) Not quite the right mixture: Chamorro and Malti as candidates for the status of mixed language. In Matras and Bakker (eds.), The Mixed Language Debate: Theoretical and Empirical Advances . Berlin: Mouton de Gruyter, pp. 271-316. Thomason, Sarah G. and Terrence S. Kaufman, 1988, Language Contact, Creolization, and Genetic Linguistics. Berkely, Cal.: University of California Press. Wartburg, Walter von (1939). Réponses au Questionnaire du Ve Congrès international des Linguistes.
Winford, Donald (2003) An introduction to contact linguistics . Oxford: Blackwell.
