On a previous post I commented about how the major language families of South America appear to form two blocks based on shared vocabulary and morphology. Here I would like to elaborate a bit more on that, recognising that this “lumping” tendency is actually nothing new but pays tribute to the ideas of Greenberg and Ruhlen.
I previously presented some basic vocabulary in different language families of South America, noticing some similarities. Let’s see just one of them, the word for tongue:
‘tongue’ in many tongues
Most of the ‘Andean’ (for lack of better term) languages make use of a root similar to *kal ~ *lak or anything like that (about the feasibility of this, check Latin lac, lactis and Greek γάλα, γάλακτος, both derived from PIE *glákt-). In contrast, the ‘Amazonian’ languages tend to prefer some version of *(n)VnV or similar forms. The words in the two groups above are actually listed under different entries in Greenberg and Ruhlen’s etymological dictionary, but I suppose they might as well be grouped together (I’m not taking this too seriously as an application of historical linguistics, by the way; this is just an exercise to think about the big picture).
I recognise that many of those resemblances could be due to some universal associations between sound and meaning. However, I would still prefer to see the similarities within each continent or region as being due to historical relationships. If there were resemblances between those words in some Amerind languages and, let’s say, a lot of Australian aboriginal ones, then I suppose invoking universals would be appropriate.
Now, as we can see from the table above, there are some exceptions in each group. The languages of the Guaykuru family, mostly spoken in the Paraguayan Chaco, follow the ‘Andean’ pattern, whereas Huarpe, spoken in Patagonia, shares the ‘Amazonian’ cognate.
Another important aspect is the possessive pronouns. They tend to be suffixes in Andean languages like Quechua and Aymara, but prefixes in languages of the lowlands (as I commented in the previous post, this sets Puquina apart from its neighbours and brings it closer to the Arawak family).
I believe the most important trait, however, appears in verb morphology. I have been thinking that South American languages follow two main patterns. Languages of the western part of the continent, mainly those spoken in the Andes (Quechua, Aymara) and adjacent Pacific coast or eastern slopes (Muchik, Mapudungun, Jivaroan) tend to use a lot more (if not exclusively) suffixes to mark agents, patients, tenses and aspects of the verb. In that way, they are similar to most Eurasian languages.
On the other hand, languages spoken east of the Andes, including the Amazonian families (Arawak, Tupi, Karib) as well as those of the Chaco and Cerrado (Macro-Je, Guaykuru) tend to use prefixes to mark the persons in the verbs. There are a number of isolating languages in the continent, but because most American languages tend to be polysynthetic (i.e. adding a lot of affixes to create more complex sentences from a single root), these two patterns are easily identifiable.
In the example above, I highlighted all the suffixes in blue and all the prefixes to the verbal root in red. The languages of the ‘Andean’ sphere don’t seem to make use of prefixes. Those of the ‘Amazonian’ world do, mostly for marking the persons, but they also rely on suffixes for the tense, aspect, mode of the verbs. The Nambikwaran languages are outliers, being heavily suffixing. Another obvious exception is Arawak. I still think it should be included in the second group, given the fact that agents are prefixed to the verb and possessive pronouns are prefixed to the nouns, but admittedly it does tend to use more suffixes than other Amazonian languages. By the way, if you are interested in this type of analysis, you should check the World Atlas of Linguistic Structures, especially the chapters by Johanna Nichols, where some insightful thoughts about the global distribution of such traits can be found.
Furthermore, in terms of phonology there is also a division between the ‘western’ and the ‘eastern’ blocks. Andean-Patagonian languages tend to have a lot more consonants, including sounds like /l/, /χ/, /q/ and others that are rare in the tropics. In opposition, the Amazonian languages are richer in vowels,with sounds like /
i/ and many nasal vowels. Interesting enough, as I was writing this post, I came across a paper suggesting that drier, colder climates favour the development of consonant-rich languages, whereas the humid tropics favour vowel-rich ones (that has to do with the way sound dissipates). This goes along similar lines as the theory about the development of tone in warm humid climates that I mentioned before. Whether we agree with that or not, it’s at least an interesting hypothesis.
The division above is not too different from the original proposal of Joseph Greenberg or, perhaps even better, the revised proposal with Mehrit Ruhlen. The original distinction was between a Macro-Chibchan, an Andean-Equatorial, and a Ge-Pano-Carib group. The Andean-Equatorial group, however, included Arawak and Tupi, families that I see as being typologically very different form the Andean ones. Later on, in the etymological dictionary, he and Ruhlen slightly changed the classification, grouping the South American families into Andean-Chibchan-Paezan, Equatorial-Tucanoan, and Ge-Pano-Carib. I think this last proposal is quite elegant and broadly corresponds to the binary division that I suggested above (see the figure below). Most importantly, as I will explain in the remainder of this post, I believe a division along the lines of Andes-Patagonia vs. Amazonia-Central Brazil has intriguing archaeological correlates.
First, it is possible that the division is rooted in the deep past, dating to the time of initial colonisation of the continent. This is too much of a controversy to be addressed here, but, as I said in the previous post, there is enough evidence for human presence since at least 18 to 14 thousand years ago in South America. Many early sites are found along the Pacific coast and adjacencies, like the Paijan complex of the central coast of Peru, or the Monte Verde site of Chile. Further south, at the tip of the continent, we have the Fishtail horizon of Tierra del Fuego. Genetically, there is evidence that a Pacific coastal route was utilised during the initial human migration into the Americas, with South American peoples living in the western part of the continent still exhibiting a very early founder haplogroup.
Despite the diversity of archaeological cultures in South America during the late Pleistocene, there seems to be a division between projectile points (like those of Taima-Taima, Paijan, Tierra del Fuego) and a different type of stone tool mainly found in Brazil – the unifacial tools. Instead of being worked bifacially, i.e. on both sides like a Clovis arrowhead, these are worked on only one of the faces, resulting in plano-convex artefacts commonly referred to as ‘scrapers’. The best example is the Itaparica culture of the Central Brazilian savannas, dating to around 12 thousand years ago, but this unifacial tradition may be extended further back in time if the dates for northeastern Brazil are confirmed. It is tantalising that there might have been an early migration to the Americas later superseded by the modern Amerind populations, the genetic evidence of which is now found in Amazonia (just to play the Devil’s advocate, this has also been disputed).
In summary, I would suggest that the big Andes-Amazonia linguistic divide seen in vocabulary and morphology is explained by the deep histories of human migration in South America. Ancient population waves took separate paths, one along the Pacific and adjacent Andes, the other penetrating the tropical lowlands.
To conclude, I would like to point out that, at the interface between the two groups, we will find many examples of contact between Highland and Lowland languages. Many Amazonian languages of the eastern slopes of the Andes (like Jivaro) show a typically Andean pattern. Maybe that explains the heavy use of suffixes in Arawak languages, as a western Amazonian origin for that widely diffused family is becoming more accepted. Even more intriguing is the case of the Pano languages (see examples above).
I always thought that they somehow ‘sounded’ Andean, at least in their use of suffixes, like the -miski that is so frequent in Cashinahua. The negative suffix -ama- in the same language even has a Quechua counterpart. The relationship between the Pano and the Andean sphere had already been noticed by archaeologist Donald Lathrap in a paper that is quite difficult to find. He advocated that the ceramics of the Cumancaya tradition, appearing in the Ucayali around AD 800, have an iconography inspired by ceramics of Highland Ecuador. He believed they were introduced in the Amazon by a foreign elite during the middle Horizon, when the lowlands traded with the Wari empire. In his classic ‘the Upper Amazon‘, Lathrap brought that interaction further back in time, noticing some iconographic resemblances between the pan-Andean Chavín style (900-200 BC) and the Shakimu ceramics of the Ucayali (650 BC).