Linguistweets Conference (ABRALIN)
This page contains the Twitter presentation and additional comments to it with references. The content was presented (tweeted) by Fabrício Ferraz Gerardi, Tiago Tresoldi and Stanislav Reichert on the 05 of December 2020. This is ongoing work in the initial phase.
Tupí-Guaraní (TG) is the largest linguistic family of South America. We want to know more about its spread (peoples and languages). Historical linguistics has tools to investigate this issue aided by digital data, computers, and methods from evolutionary biology.
| Tupí-Guaraní Languages | Phylogenetic Tree |
|---|---|
 |
 |
See Nunn (2011).
Using phylogenetic tools (BEAST2, beastling), we build linguistic trees from open-access reusable data in CLDF, lifted with EDICTOR from TuLeD (285 concepts, 16211 words, 41 TG langs of 78 in TuLeD). 2832 cognate sets detected with LexStat, 40% manually reviewed (ongoing work).
| TuLeD | CLDF | EDICTOR |
|---|---|---|
 |
 |
 |
EDICTOR is an excelent tool for historical linguists to process their data, visualize multiple lines of alignments and segmentations.
CLDF, it is a standard for cross-linguistic data formats, allowing FAIR data in Linguistics.
Linguistic trees display classifications comparable with results from other fields (Archaeology, Ethnography, History). We test a model on lexical data only, then we interpret results considering extralinguistic data, evaluating hypotheses, and improve data and model accordingly.
| Archaeology | Archaeology | Ethnographic Information | Amount of cognates |
|---|---|---|---|
.png) |
 |
 |
 |
By interpreting results considering extra-linguistic data we mean looking at the trees and interpreting them also considering, e.g., archeological or ethnographical data. The case of Ka'apor is illustrative. Ka'apor seems to have had a relatively long contact with either Tupinambá or Língua Geral (cf. Ribeiro 1996; Correa da Silva 2011), nonetheless, based on phonological criteria, Ka'apor is grouped with Wayampi, Emerillon, Anambé of Ehrenreich, and Guajá, languages by Rodriges and Cabral (2011) that are today far apart. In spite of the position of Ka'apor in the tree, we have linguistic data (Rodrigues and Cabral 2011), and the myths of the Ka'apor giving clues to their location more to the west in the past (cf. Huxley 1963; Ribeiro 1996; Balée 1994).
Evolutionary methods: we start from simple (NJ) to highly complex ones for millions of trees (Bayesian MCMC, covarion model, relaxed clock, independent rates of variation). Basic calibrations from literature. Challenge: deliver the best single tree that summarizes the results.
| Neighbor Joining | NeighbourNet | Density Tree |
|---|---|---|
 |
 |
 |
Neighbor Joining (NJ) is the clustering algorithm most frequently used in phylogenetics, which does not require an evolutionary clock. Neighbor joining is more of a quick method of approximating the minimum evolution tree, usually with a confidence inversely proportional to the distance from the root (i.e., it tends to identify recent splits better than older ones). It is a common first approach for data exploration and is expected to be not too different from a Bayesian one.
A NeighbourNet (NN) is a kind of split graph that is commonly used to visualize conflicting signals, such as those resulting from lexical borrowings. It is a quick method for identifying reticulate relationships and, while it does not group languages in terms of genealogy, can be used to visually estimate splits and rates of variation and non-overlapping evolutionary trajectories. The one here reported has a delta score of 0.3972 and a q-residual score of 0.006793.
The Density Tree plots thousands of the best scoring trees on top of each other (in green), highlighting conflicting signals that can be due to a number of reasons, from limited data to a high ratio of borrowings. The summary of the best trees, trees with higher probabilities, is plotted in blue. It also allows us to visually assess how likely some splits and their dates are. In our preliminary density tree, above, where the "root canal" is highlighted in blue, we can see that the groups for Guarani and Tupinambá (Ferraz Gerardi and Reichert 2020) seem clearly distinct, with some probability that it happened earlier than the date of root canal. Some languages also show some conflicting signal: in this analysis, for example, Old Guarani is closer to the Siriono group, but there is a strong signal (visible by the green shade) for a closer membership with the Mbya group. The positions of Zo'e and Kamajurá, as expected, also show a lower degree of confidence.
Mawe-Awetí-TG hypothesis recognized. Fast expansion of Guaraní (light blue) until 400 years ago. 4 major groups identified. Blue group supports ethnogr/hist/ling evidence: dialect continuum, common area of origin, close contact. Fits archaeological dates (Almeida & Neves 2015).
| Mawetí-Guaraní Hypothesis | Phylogenetic Tree |
|---|---|
 |
 |
The phylogenetic tree above is a consensus tree, computed with a 20% burn-in, which is the best preliminary results so far, accepting some of the monophyletic restrictions given in the literature and confirmed in our previous analyses. Sateré-Mawé forms the first out-group, followed by Awetí, with full confidence (the posterior support at 1.00 for both) (Silva 2011, Meira & Drude 2015). The groups highlighted by the root canal in the density tree are confirmed and better organized, with some lower support on the initial moments of expansion. These lower support might be improved with more data and longer analyses, but might also support hypotheses on a rapid expansion in different groups. As for dates, this consensus tree suggests that the expansion of the group, after the ancestor of Awetí had already split, started around 1,100 years ago, with a rapid speciation until circa 500 years ago. The less resolved relationship is between the various Guarani languages, which might indicate a continous contact. The horizontal axis represent time in a scale of 100 years.
Collect more historical info, improve cognacy. Cognacy alone is not all; include other linguistic data. Constrain and calibrate model(s) (dates, geography, ratios, monophyletic groups). More info, references, and comments: https://tular.org/tgtweet. Thank you!
Almeida, F. O. D., & Neves, E. G. (2015). Evidências arqueológicas para a origem dos Tupi-Guarani no leste da Amazônia. Mana, 21(3), 499-525.
Balée, W. L. (1994). Footprints of the forest: Ka'apor ethnobotany-the historical ecology of plant utilization by an Amazonian people. Columbia University Press.
Bouckaert R., Heled J., Kühnert D., Vaughan T., Wu C. H.,Xie D., et al. (2014). BEAST2: A Software Platform for Bayesian Evolutionary Analysis. PLoS Comput Biol. 10(4).
Ferraz Gerardi, F & Reichert, S. (2020). The Tupi-Guaraní language family: a phylogenetic classification. To appear in Diachronica.
Greenhill, S. J., Heggarty, P., & Gray, R. D. (2020). Bayesian Phylolinguistic. In R. D. Janda, B. D. Joseph, & B. S. Vance (Eds.), The Handbook of Historical Linguistics (pp. 226-253). Hoboken, New Jersey: Wiley-Blackwell.
Huxley, F. (1963). Selvagens amáveis:(um antropologista entre os índios Urubus do Brasil). Brasiliana.
Maurits, L., Forkel, R., Kaiping, G. A., Atkinson, Q. D. (2017) BEASTling: A software tool for linguistic phylogenetics using BEAST 2. PLOS ONE.
Meira, S., & Drude, S. (2015). A summary reconstruction of Proto-Maweti-Guarani segmental phonology. Boletim do Museu Paraense Emílio Goeldi. Ciências Humanas, 10(2), 275-296.
Nunn, C. L. (2011). The comparative approach in evolutionary anthropology and biology. University of Chicago Press.
Rodrigues, A. D., & Cabral, A. S. A. C. (2002). Revendo a classificação interna da família Tupí-Guaraní. Línguas Indígenas Brasileiras. Fonologia, Gramática e História, Atas do I Encontro Internacional do GTLI da ANPOLL, 1.
Rodrigues, A., & Cabral, A. S. (2012). Tupían. In L. Campbell & V. Grondona. The Indigenous Languages of South America: a comprehensive guide.
Rodrigues, A. D., & Dietrich, W. (1997). On the linguistic relationship between Mawé and Tupí-Guaraní. Diachronica, 14(2), 265-304.
Silva, Beatriz Carretta C. D. (2011). Mawé/Awetí/Tupí-Guaraní: relações lingüísticas e implicações históricas. University of Brasília. Unpublished PhD thesis.
Wilkinson, Mark D.; Dumontier, Michel; Aalbersberg, IJsbrand Jan; Appleton, Gabrielle; et al. (15 March 2016). "The FAIR Guiding Principles for scientific data management and stewardship". Scientific Data. 3: 160018. doi:10.1038/sdata.2016.18.