From explicit to implicit: constraints and explanations

This workshop seeks to deepen our understanding of linguistic flexibility in communication, focusing particularly on optional linguistic formulations — instances where explicit and implicit forms of a linguistic structure convey essentially the same meaning. Some examples are: • optional discourse connectives: Such as “because” and “but” which can mark explicitly coherence relations at the discourse level (or leave them implicit without such a connective). E.g., He studied hard at the exam, (but)he still failed, unfortunately. • variations in specificity: The choice between expressions with varying degrees of specificity, e.g., I walked (vs. went) to the store. or the use of an ambiguous rather than specific connective, e.g., He worked very late, and (instead of ‘but’) he was already up at dawn. • argument drop (pro-drop): Particularly common in languages like Chinese, Japanese, and Portuguese, where both subject and object pronoun can be omitted when recoverable from context. For example, in Chinese: A: 你喜欢这部电影吗？ (Do you like this movie?) B: —不喜欢—. (I do Not like it.) • topic drop: In German, for instance, subjects or objects can be omitted only when they are in the sentence initial (topic) position (Huang and Yang, 2024), e.g., — habe ihn gestern schon gesehen — have him yesterday already seen — habe ich gestern schon gesehen — have I yesterday already seen *Ich habe — gestern schon gesehen *I have — yesterday already seen *Ihn habe — gestern schon gesehen *him have — yesterday already seen 2 • functional words: Such as “that” “where” and “which” introducing complement and relative clauses. E.g., This is the place (where) we met for the first time. • bare infinitives vs. infinitives with "to": e.g., “help do” vs. “help to do.” Such flexibility in language use raises intriguing questions about the factors that drive speakers and writers to choose one form over another and the impacts these choices have on the process and the effectiveness of communication. In this workshop, we aim to explore these issues from a comprehensive point of view, breaking them down into specific aspects that can be addressed through various theoretical frameworks and hypotheses. Especially today, with the growing advantage of multi-/inter-disciplinary research methodologies and the increasing availability of diverse datasets, we expect to gain even deeper and richer understanding of the mechanisms behind these linguistic behaviors. Below, we outline the key questions and perspectives we aim to address, with either a focus on an individual language or a cross-linguistic approach. These serve as the foundation for our discussions while remaining open to broader contributions and innovative ideas. • Medium and Genre Sensitivity: Are certain forms preferred systematically in written, spoken, or computer-mediated communication, given the differences in time, space and immediacy between writer/speaker and reader/listener (Biber, 1986; Chafe, 1982; Herring, 2011; Xiao et al. 2021)? How do genre and medium conventions influence linguistic choices? For instance, scientific discourse in English is found to vary in the overall lexical and verbal density in research articles (61%) and oral conference presentations (46%) (Hamilton and Carter-Thomas, 2017). • Diachronic Trends: (How) Do linguistic choices evolve over time, especially in response to the prevalence of digital communication, which brings new genre features regarding language use at multiple levels (Crystal, 2006; Leppänen et al., 2017)? • Cognitive Constraints: (How) Do cognitive biases, such as the tendency to read consecutive propositions in a text as temporarily continuous or causally related, influence the linguistic formulations of continuous (e.g., causal, additive) and discontinuous (e.g., adversative, concessive) relations? And how do the variant formulations, such as implicit vs explicit signaling of these relations, influence the online processing of such connections, given the default assumptions (Levinson, 2000; Murrey, 1997; Sanders, 2005)? • Typological Differences: (How) Do typological differences impact language structure and use? For example, pro-drop in German is limited to the sentence initial positions, whereas Chinese and Japanese permit more extensive pro-drop (Huang and Yang, 2024; Schäfer, 2021). Scientific discourse in French displays greater lexical variation, whereas English tends to pack 3 information more densely with frequent reuse of the same lexical items (Hamilton and Carter-Thomas, 2017). Additionally, the causality-by-default hypothesis, which suggests that causal coherence relations are often left implicit without a connective, is found less applicable to Turkish and Lithuanian, wherein causal relations are more frequently and explicitly marked with causal connectives (Mendes et al. 2023). • Audience Design vs. Production Economy: Is the selection of linguistic form listener-oriented, attending to listeners’ interpretive needs and cognitive load, or is it more speaker-oriented, prioritizing ease and efficiency in production (Gahl et al., 2012; Goodman and Frank, 2016; Yung et al., 2021)? • Information Density: How does the distribution of information density across an utterance influence ellipsis or “redundancy”? For instance, the relational information between consecutive clauses may be encoded via multiple devices, not only the prototypical discourse connectives but also alternative lexical and syntactic signals. How do these elements balance over the whole utterance to achieve a uniform information density as an optimal strategy for communication (Asr and Demberg, 2015; Crible and Demberg, 2020; Egg and Das, 2022; Jaeger, 2010; Levy and Jaeger, 2006; Schäfer et al., 2021)? The objective of this workshop is to bring together researchers exploring these and related phenomena and questions, from a variety of theoretical frameworks and using diverse research methods, such as corpus analysis, experimentation, computational modeling, and NLP approaches. We look forward to collaborative and dynamic discussions, which we believe will bring fresh and/or complementing insights, especially regarding inconsistencies or partial confirmations found in the literature. These findings suggest complex interplay of factors such as cognitive bias, genre, medium (written, oral, digital), as well as typology of language. Together, we hope to deepen our understanding of the constraints underlying the selection of optional linguistic formulations as well as their impact on language production, processing, and comprehension.