General Linguistics Colloquium (WiSe 2022/2023)

In this talk, I will present and discuss the results from my recent study on linguistic variation inside the city of Buenos Aires (Kellert & Matlis 2022). As will be shown measuring linguistic variation on these scales is challenging due to the lack of highly-spatially-resolved data as well as to the daily movement or users’ “mobility” inside cities, which can obscure the relation between the social context and linguistic variation. I will demonstrate how combining geolocation with user IDs and textual analysis of tweets can yield information about the linguistic profiles of the users, the social context associated with specific locations and their connection to linguistic variation.
Kellert Olga & Nicholas Matlis 2022 ‘Social context and user profiles of linguistic variation on a micro scale.’ Proceedings of the Ninth Workshop on NLP for Similar Languages, Varieties and Dialects, pages 14–19, October 16, 2022. ©2022 Association for Computational Linguistics, https://aclanthology.org/2022.vardial-1.2.pdf

In this talk I will present my ongoing study, the aim of which is to present the vocalic systems of Ancient Greek dialects of ca. 6th century BCE using Optimality Theory (and its variants when necessary) as framework. Here I will present the results of my work on the originating from Common Greek long /a/, /æ/ and /ɛ/ in Attic, East Ionic and Laconic Doric Greek and their likely underlying representations. Firstly I will present the distribution of these phonemes in the dialects in question; secondly, the constraints necessary for the correct evaluation of the given inputs, and their hierarchies in these dialects, and finally the evaluation itself. In the end I will discuss the future plans and hopes for this project.

During this talk, I will not present the results and definitive conclusions from a study that has already been completed. Instead, I will use the opportunity to present a selection of running projects. I will present to you ideas at different stages of development, all focussed on understanding how our decisions are influenced by the typological properties of the language we talk, by the society we live in, by cultural values we ascribe to, and how these factors interact with each other affecting personal experience and judgements at the individual level, and also the way of reasoning in different populations at the group level.
Talking also on behalf of my collaborator, I will try those of you who are thinking about research career at the PhD level or junior postdoc levels get in touch and discuss joining us on these projects, supported by international foundations (e.g., European Research Council), national foundations (the Alexander von Humboldt Stiftung, the Portuguese Foundation for Science and Technology, the Spanish Ministry for Science and Innovations), and by industrial partners (mostly by pharmaceutical and biomedical companies).

In this talk, I will present ongoing joint work on analyzing metaphor in religious language. Metaphor, defined as a "mapping across two conceptual domains" is an ubiquitous phenomenon in language. It fulfills a special role in religious language in that it enables us to express ideas about an abstract entity by referring to a well-known concrete entity and, in this way, to make statements about the transcendent.

In the talk, I first present our work on annotating deliberate metaphors following the Five Step Method proposed by Steen (2007) and discuss the analysis of selected complex cases. Second, I report on a pilot study in which we are applying Shutova et al's (2013) approach to German-language religious texts and present first results from this study.

The recent advances in technology, data science, and artificial intelligence pushed a computational turn in the humanities and linguistics, where the role of NLP is fundamental to support the processing, the interpretation and the analysis of textual data. More and more studies are devoted to exploit NLP techniques for linguistic studies traditionally addressed by human-intensive processes of "close-reading". For example, the use of word embeddings proved to be an effective mean in the field of semantic change or Semantic Shift Detection. Goal of Semantic Shift Detection is to detect, interpret, and assess changes of word meanings over time. The aim of this talk is to present the hypotheses and implications of recent word embedding techniques and to show an example of their application.

In his discussion of enclitic pronouns Berthold Delbrück states that “[d]ie Enclitica wird [sic!] von dem am stärksten betonten Worte, und das ist das erste im Satze, wie von einem Magnet angezogen” (1878: 48, cf. Delbrück 1888: 33). For the majority of Vedic particles he comes to a similar conclusion. Thus, on ha he notes: “[d]emnach darf man sagen, dass ha eine das erste Wort hervorhebende Partikel sei…” (1888: 501). In current studies on Vedic and Indo-European syntax there is general agreement that these pronouns and particles are second position or Wackernagel clitics (see among many others Hale 1987; Hock 1996; Krisch 1990; Lowe 2011). However, while Delbrück’s observations undoubtedly hold true for Vedic prose, the picture in the Rigveda is actually much more diverse. Wackernagel clitics may even surface in sentence- and verse-final position: (1) imám yajñám nayata devátā naḥ Lead this sacrifice of ours to the gods! (2) uruṣyátaṃ jaritā́raṃ yuváṃ ha Protect the singer, the two of you! (RV 4.58.10) (RV 4.43.7) Placement in second position outweighs other positions in the sentence by far, but deviations from the expected pattern as in exx. (1) and (2) are far too numerous and far too systematic to be dismissed offhand. It cannot be excluded that the difference between the pattern observed by Delbrück for Vedic prose and the one attested in the Rigveda is the result of diachronic developments. However, it is more likely that the striking variation attested in the Rigveda is connected to its genre. But while poetic license may be considered a necessary cause for the observed patterns, it is by no means sufficient: As Jakobson pointed out in several influential papers (e.g. Jakobson & Bogatyrev 1929), poetic speech is confined by grammatical constraints, even if the poets stretch acceptability to its limits. The attested aberrant patterns should therefore be considered grammatical. As a consequence, they must be accounted for by any model designed to capture second position phenomena in Vedic (for a previous attempt see Hale 1995). The first part of this talk is based on a corpus study of the 1.pl pronoun naḥ in the Rigveda. The clitic naḥ is attested 2408× (Lubotsky 1997). While the majority of cases is actually 2P, most of the deviant cases follow clear patterns: (1) There is a strong tendency to place the clitic adjacent to its head, even if it is hosted by another constituent. (2) In many attestations the clitic is still in second position, albeit within a domain smaller than the sentence. Most frequently, this domain is what can be described as the verbal complex. In this respect, Early Vedic patterns with languages like Old Lithuanian (Petit 2010; Razanovaitė 2014) and Gothic (Miller 2019). In the second part of the talk an attempt is made at extending the model for the placement of 2P clitics in Vedic developed in Keydana (2011) so as to capture the complexity of patterns actually observed in the corpus.