Mutual intelligibility and surprisal in Slavic intercomprehension

SFB 1102 (C4) | Web experiments  | Project wiki

he project INCOMSLAV investigates the relation between information density, encoding density and grammaticalisation in a cross-linguistic perspective, focusing on intercomprehension within the family of Slavic languages. In the initial funding period (2014-2018), the project brings together results from the analysis of parallel corpora and from a variety of experiments with native speakers of Slavic languages and compares them with insights of comparative historical linguistics on the relationship between Slavic languages. A statistical language model of surprisal is used to measure information density and as a tool to gauge how language users master high degrees of surprisal, due to partial incomprehensibility. The key idea here is that comprehension of an unknown, but related, language should be better, when the language model adapted for understanding the unknown language exhibits relatively low average surprisal, or density. In the second funding period (2018-2022), the research agenda is extended to spoken language, which allows us to investigate how information density is balanced between the acoustic and the text level in successful intercomprehension. At all levels from the acoustic signal and its phonetic structure to the texts generated from speech we develop similarity metrics and information density measures related to Slavic intercomprehension.

In the first two phases of the CRC, the empirical focus of C4 was on the mutual intelligibility of visual (written) or auditory (spoken) input for speakers of closely related languages in the Slavic language family. Experimental and modelling work in the second phase, which has combined methods from language, speech and translation technology, has provided a wealth of findings highlighting how information density is distributed across the acoustic and the text channels in successful intercomprehension. Based on these results, we are now in a position to address, in the third phase (2022-2026), core properties of intercomprehension as they unfold in goal-oriented communication, characterized by cooperative behaviour and adaptive interaction. This overarching goal entails the investigation of linguistic structures beyond lexical similarity and word sequence based predictability, taking into account constructional similarity, the crosslingual transparency of multi-component units, and prosody. Specifically, conversational dialogue-style experimental setups are employed in order to explore the (ex)change of information as the interaction unfolds. We will develop models of surprisal capturing the information conveyed by multi-component units and prosodic features, in particular intonation. Finally, C4 will validate the scalability of our results and models in terms of a transfer to a selected set of features of other language families, e.g. Semitic.

PhD research staff (phase 1): Andrea Fischer, Klára Jágrová, Irina Stenger

PhD research staff (phase 2): Yu Tracy Chen, Badr Abdullah, Jacek Kudera

PhD research staff (phase 3): Iiuliia Zaitova, Badr Abdullah (Postdoc)

Recent pblications & Resources

...

2020

2019

Avgustinova (2019) Gegenseitige Verstehbarkeit und Surprisal in Slavischer Interkomprehension: empirische Basis und linguistische Modellierung. Invited lecture at University of Hamburg

Jagrova, Stenger, Avgustinova (2019) Slavic Intercomprehension Matrix. 13.Deutscher Slavistentag, Internationaler Kongress der deutschsprachigen Slavistik. Sektion: Didaktik der slavischen Sprachen und Kulturen

Avgustinova, Iomdin (2019) Towards a Typology of Microsyntactic Constructions. In: G.Corpas-Pastor, R.Mitkov (Eds.) Computational and Corpus-Based Phraseology. Springer, Cham:15-30

2018

Jágrová: Processing effort of Polish NPs for Czech readers  – A+N vs. N+A.In: Guz, Szymanek (eds.): Canonical and Non-Canonical Structures in Polish. Studies in Linguistics and Methodology vol. 12. Wydawnictwo KUL, pp. 123-143. Preprint

Jágrová, Avgustinova, Stenger, Fischer: Language models, surprisal and fantasy in Slavic intercomprehension, Computer Speech & Language, Available online 12 June 2018, ISSN 0885-2308, https://doi.org/10.1016/j.csl.2018.04.005.

2017

Jágrová, Stenger, Marti, Avgustinova. (2017). Lexical and Orthographic Distances between Czech, Polish, Russian, and Bulgarian - a Comparative Analysis of the Most Frequent Nouns. In: Language  Use  and  Linguistic Structure.  Olomouc  Modern  Language  Series,  Palacký University Olomouc. pp. 401-416 (online)

2016

2015