Sebastian Drude

Genealogical relations and lexical distances within the Tupian linguistic family

In this paper we present the first results of the application of computational methods, inspired by the ideas in McMahon & McMahon (2005), to a dataset collected from languages of every branch of the Tupian family (including all living non-Tupi-Guarani languages) in order to produce a classification of the family based on lexical distance. We used both a Swadesh list (with historically stabler terms) and a list of animal and plant names for results comparison. In addition, we also selected more (HiHi) and less (LoLo) stable terms from the Swadesh list to form sublists for indepedent treatment. We compared the resulting NeighborNet networks and neighbor-joining cladograms and drew conclusions about their significance for the current understanding of the classification of Tupian languages. One important result is the lack of support for the currently discussed idea of an Eastern-Western division within Tupi.

Motion history images for online speaker/signer diarization

We present a solution to the problem of online speaker/signer diarization - the task of determining who spoke/signed when?. Our solution is based on the idea that gestural activity (hands and body movement) is highly correlated with uttering activity. This correlation is necessarily true for sign languages and mostly true for spoken languages. The novel part of our solution is the use of motion history images (MHI) as a likelihood measure for probabilistically detecting uttering activities. MHI is an efficient representation of where and how motion occurred for a fixed period of time. We conducted experiments on 4.9 hours of the AMI meeting data and 1.4 hours of sign language dataset (Kata Kolok data). The best performance obtained is 15.70% for sign language and 31.90% for spoken language (measurements are in DER). These results show that our solution is applicable in real-world applications like video conferences and information retrieval.

Unsupervised Feature Learning for Visual Sign Language Identification

Prior research on language identification focused primarily on text and speech. In this paper, we focus on the visual modality and present a method for identifying sign languages solely from short video samples. The method is trained on unlabelled video data (unsupervised feature learning) and using these features, it is trained to discriminate between six sign languages (supervised learning). We ran experiments on short video samples involving 30 signers (about 6 hours in total). Using leave-one-signer-out cross-validation, our evaluation shows an average best accuracy of 84%. Given that sign languages are underresourced, unsupervised feature learning techniques are the right tools and our results indicate that this is realistic for sign language identification.

Variation in Tupi languages: Genealogy, language change, and typology

At least 40 spoken languages form the large tupi family in its subfamilies tupi-Guarani, Mawe, Aweti, Arikem, Juruna, Monde, tupari, Munduruku, ramarama and Purubora, providing a wealth of data for linguistic studies about variation – variation explained by genetic relations (common origin, ultimately from the presumed language ‘proto-tupi’) or by contact relations with other indigenous or non-indigenous languages. the interest in indigenous languages has increased in recent years, evinced by the publication of a growing number of descriptive, historical-comparative and other studies. Most studies published in this volume originate from a linguistic symposium organized by Wolf dietrich and sebastian drude at the 54th international Congress of Americanists in Vienna, in 2012. the symposium was dedicated to “Historical variation and variation by contact among the tupian languages”. those studies that deal with problems of linguistic genealogy, genetics, language change, and syntactic typology across several tupi languages or a single tupi language are published in this special “dossier”. the first six papers deal with problems of the whole tupi family or at least one of its sub-families. three of them investigate evolutionary topics; three are cross-linguistic synchronic studies. Eduardo dos santos and colleagues from the area of human genetics, in their paper “origins and demographic dynamics of tupi expansion: a genetic tale”, use recent genetic data in order to show that the Madeira-Guapore region may indeed be considered to be the tupi homeland. Ancient tupi expansion within the Madeira-Guapore region and dispersion to other south American areas seems to be related to patrilocal practices. this outcome allows for new interpretations of archaeological and linguistic data, for instance the dispersion of female associated technologies like ceramics and terminologies related to ceramics. Ana Vilacy Galucio and colleagues from the tupi Comparative Project, in their paper “on the genetic relationship and degree of relatedness with the tupi linguistic family”, present the first lexicostatistical and phylogenetic attempt of the genetic classification of most languages of the tupi family, including four languages of the tupi-Guarani branch. Based on all relevant previous studies of particular branches of the family, and applying lexicostatistics to a semantically based word list,, the article demonstrates that the two major branches of the state of rondonia, Monde and tupari, have high percentages of shared cognates. this supports the results of the first article of this volume. the paper by sergio Meira and sebastian drude gives an overview of their reconstruction of the segmental phonology of Proto-Maweti-Guarani, the hypothetical proto-language from which modern Mawe, Aweti, and the

Reduplication as a Tool for Morphological and Phonological Analysis in Awetí

This chapter describes reduplication in verb forms of Aweti, a Brazilian indigenous language spoken by a small group living in three villages in the region of the headwaters of the Xingu River in Mato Grosso, central Brazil. The study of reduplication can contribute to the understanding of phonological and morphological units and processes. The chapter summarizes the phonological system of Aweti and presents some basic morphological properties and some diagnostic affixes used in the remaining sections. It gives a short general overview of verbal reduplication in Aweti and describes the basic patterns of reduplication found with active (transitive and intransitive) verbs. The chapter uses evidence from reduplication for a closer look at certain (morpho-)phonological alternations at the beginning and end of verbal stems. It treats the reduplication with stative verbs. The chapter discusses the domain of reduplication if derivational affixes occur, and summarizes the most important findings. Keywords: active transitive verbs; Aweti; morpho- phonological alternations; stative verbs; verbal reduplication

Application of audio and video processing methods for language research and documentation: The AVATecH Project

Evolution and changes of all modern languages is a well-known fact. However, recently it is reaching dynamics never seen before, which results in loss of the vast amount of information encoded in every language. In order to preserve such rich heritage, and to carry out linguistic research, properly annotated recordings of world languages are necessary. Since creating those annotations is a very laborious task, reaching times 100 longer than the length of the annotated media, innovative video processing algorithms are needed, in order to improve the efficiency and quality of annotation process. This is the scope of the AVATecH project presented in this article.