Special Issue

Special Issue in the Journal ‘Linguistics’ accepted

Effects of Context on Language: An Information-theoretic Perspective


Editors: Stefania Degaetano-Ortlieb, Ekaterina Lapshinova-Koltunski, Marie-Pauline Krielke, Heike Przybyl

    The proposed special issue addresses a central topic in the field of linguistics by examining the complex interplay between contextual factors and language processing. As language is a core aspect of human cognition and communication, understanding the role of context in shaping linguistic behavior is crucial for advancing knowledge in the discipline. We think that this issue will be highly relevant to the readership of ‘Linguistics’.

    The proposed special issue addresses a central topic in the field of linguistics by examining the complex interplay between contextual factors and language processing. We adopt an information-theoretic perspective as it allows us to capture the impact of contextual variance on language processing. As language is a core aspect of human cognition and communication, understanding the role of context in shaping linguistic behavior is crucial for advancing knowledge in the discipline.
    The study of context in language has long been acknowledged as essential for understanding language production and comprehension such as Grice’s (1975) cooperative principle and his maxims of conversation or Sperber and Wilson’s (1986) relevance theory. These theoretical frameworks emphasize the importance of shared background knowledge, expectations, and inferential processes in driving effective communication. In this special issue, we conceptualize context not as a static entity, but as a dynamic construct influenced by various linguistic and extra-linguistic factors (e.g. situational context, timeframes). This dynamic understanding of context aligns well with the idea of an information-theoretic perspective that can capture the multifaceted nature of context showing its potential to significantly influence language production and comprehension.


    The aim of this special issue is to advance our understanding of how linguistic choices are influenced by contextual factors, adopting an information-theoretic perspective. Coming from a diverse set of disciplines (contrastive and diachronic linguistics, translation and interpreting studies, language typology, and cognitive linguistics), the contributions consider as contextual factors linguistic (ranging from local linguistic context up to clause, sentence, or text levels) and, to some extent, extra-linguistic factors (e.g., time, language, written vs. spoken mode, register, and individual differences).
    The methodological approaches include corpus-based analyses, statistical testing, experimental paradigms, and computational modeling.

    The papers contributing to this special issue share a common trait: each employ one or several measures to quantitatively assess and compare aspects of information given context. Being the most prominent information-theoretic measure, surprisal is used to estimate the amount of information given the very local linguistic context of a linguistic unit. Additionally, related measures are employed as well as compared to the established measure of surprisal. Bizzoni et al. correlate surprisal of nouns with their hierarchical lexical level (general vs. specific), comparing the two cognitively different processes of translation and interpreting. Talamo et al. use long-distance surprisal to capture cross-sentential contextual relations (reference) with preceding context up to 100 words. Wallenberg employs the Deviation of the Rolling Mean (DORM) on surprisal to investigate how surprisal evolves throughout a sentence. The concept of information weight considering given vs. new information is applied to examine how much a text or clause leans towards introducing new information vs. relying on previously established information, as shown by Speyer on clauses and Fontain & Xu on texts. Degaetano-Ortlieb et al. utilize Kullback-Leibler Divergence as an information-theoretic measure to detect and analyze periods of change over centuries, considering also changes in surprisal at the lexico-grammatical and the textual level. Häuser & Kray offer a cognitive perspective on a word’s predictability within sentences as opposed to its uncontextualized predictability. Sun et al. compute surprisal using multilingual transformer-based language models and introduce an attention-aware method for computing contextual semantic relevance, comparing both types of contextually determined predictability measures on a large multilingual eye-tracking corpus.


    The rich spectrum of contributions and encompassed information-theoretic concepts underscores the comprehensive scope of this special issue. It offers quantifiable insights into linguistically interpretable notions of information, ensuring a valuable exploration of how contextual factors intricately influence linguistic choices. This contributes to a deeper understanding of information dynamics.

    Contributors:

    • Referential choice as a feature of context (Hongying Xu and Lise Fontaine)
    • Measuring accessibility through surprisal: a cross-linguistic study of personal pronouns (Annemarie Verkerk and Luigi Talamo)
    • Hierarchical semantic relations in translation and simultaneous interpreting (Ekaterina Lapshinova-Koltunski, Yuri Bizzoni, Stefan Fischer, Christina Pollkläsener and Heike Przybyl)
    • The role of informativity for variation in integrating adverbial clauses (Augustin Speyer)
    • Register formation and communicative constraints: Modeling registerial changes in English scientific writing over 300+ years (Stefania Degaetano-Ortlieb, Marie-Pauline Krielke, Yuri Bizzoni, Isabell Landwehr and Katrin Menzel)
    • CELEX or SUBTLEX? Predicting reading times of German nouns and verbs by means of two common frequency measures (Katja Häuser and Jutta Kray)
    • Topicalization is (Still) Disappearing: information uniformity as a dimension of specialization (Joel Wallenberg)
    • Attention-aware measures of semantic relevance for predicting human reading behavior (Kun Sun, Rong Wang, and Harald Baayen)