Publications

Abstract

Humans readily engage in idle chat and heated discussions and negotiate tough joint decisions without ever having to think twice about how to keep the conversation grounded in mutual understanding. However, current attempts at identifying and assessing the conversational devices that make this possible are fragmented across disciplines and investigate single devices within single contexts. We present a comprehensive conceptual framework to investigate conversational devices, their relations, and how they adjust to contextual demands. In two corpus studies, we systematically test the role of three conversational devices: backchannels, repair, and linguistic entrainment. Contrasting affiliative and task-oriented conversations within participants, we find that conversational devices adaptively adjust to the increased need for precision in the latter: We show that low-precision devices such as backchannels are more frequent in affiliative conversations, whereas more costly but higher-precision mechanisms, such as specific repairs, are more frequent in task-oriented conversations. Further, task-oriented conversations involve higher complementarity of contributions in terms of the content and perspective: lower semantic entrainment and less frequent (but richer) lexical and syntactic entrainment. Finally, we show that the observed variations in the use of conversational devices are potentially adaptive: pairs of interlocutors that show stronger linguistic complementarity perform better across the two tasks. By combining motivated comparisons of several conversational contexts and theoretically informed computational analyses of empirical data the present work lays the foundations for a comprehensive conceptual framework for understanding the use of conversational devices in dialogue.

Abstract

Establishing and maintaining mutual understanding in everyday conversations is crucial. To do so, people employ a variety of conversational devices, such as backchannels, repair, and linguistic entrainment. Here, we explore whether the use of conversational devices might be influenced by cross-linguistic differences in the speakers’ native language, comparing two matched languages—Danish and Norwegian—differing primarily in their sound structure, with Danish being more opaque, that is, less acoustically distinguished. Across systematically manipulated conversational contexts, we find that processes supporting mutual understanding in conversations vary with external constraints: across different contexts and, crucially, across languages. In accord with our predictions, linguistic entrainment was overall higher in Danish than in Norwegian, while backchannels and repairs presented a more nuanced pattern. These findings are compatible with the hypothesis that native speakers of Danish may compensate for its opaque sound structure by adopting a top-down strategy of building more conversational redundancy through entrainment, which also might reduce the need for repairs. These results suggest that linguistic differences might be met by systematic changes in language processing and use. This paves the way for further cross-linguistic investigations and critical assessment of the interplay between cultural and linguistic factors on the one hand and conversational dynamics on the other.

Abstract

A fundamental fact about human minds is that they are never truly alone: all minds are steeped in situated interaction. That social interaction matters is recognised by any experimentalist who seeks to exclude its influence by studying individuals in isolation. On this view, interaction complicates cognition. Here we explore the more radical stance that interaction co-constitutes cognition: that we benefit from looking beyond single minds towards cognition as a process involving interacting minds. All around the cognitive sciences, there are approaches that put interaction centre stage. Their diverse and pluralistic origins may obscure the fact that collectively, they harbour insights and methods that can respecify foundational assumptions and fuel novel interdisciplinary work. What might the cognitive sciences gain from stronger interactional foundations? This represents, we believe, one of the key questions for the future. Writing as a multidisciplinary collective assembled from across the classic cognitive science hexagon and beyond, we highlight the opportunity for a figure-ground reversal that puts interaction at the heart of cognition. The interactive stance is a way of seeing that deserves to be a key part of the conceptual toolkit of cognitive scientists.

Abstract

Many of the world’s languages feature an open lexical class of ideophones, words whose marked forms and sensory meanings invite iconic associations. Ideophones (also known as mimetics or expressives) are well-known from languages in Asia, Africa and the Americas, where they often form a class on the same order of magnitude as other major word classes and take up a considerable functional load as modifying expressions or predicates. Across languages, commonalities in the morphosyntactic behaviour of ideophones can be related to their nature and origin as vocal depictions. At the same time there is ample room for linguistic diversity, raising the need for fine-grained grammatical description of ideophone systems. As vocal depictions, ideophones often form a distinct lexical stratum seemingly conjured out of thin air; but as conventionalized words, they inevitably grow roots in local linguistic systems, showing relations to adverbs, adjectives, verbs and other linguistic resources devoted to modification and predication

Abstract

No class of words has better claims to universality than interjections. At the same time, no category has more variable content than this one, traditionally the catch-all basket for linguistic items that bear a complicated relation to sentential syntax. Interjections are a mirror reflecting methodological and theoretical assumptions more than a coherent linguistic category that affords unitary treatment. This chapter focuses on linguistic items that typically function as free-standing utterances, and on some of the conceptual, methodological, and theoretical questions generated by such items. A key move is to study these items in the setting of conversational sequences, rather than from the “flatland” of sequential syntax. This makes visible how some of the most frequent interjections streamline everyday language use and scaffold complex language. Approaching interjections in terms of their sequential positions and interactional functions has the potential to reveal and explain patterns of universality and diversity in interjections.

Abstract

Large language models that exhibit instruction-following behaviour represent one of the biggest recent upheavals in conversational interfaces, a trend in large part fuelled by the release of OpenAI's ChatGPT, a proprietary large language model for text generation fine-tuned through reinforcement learning from human feedback (LLM+RLHF). We review the risks of relying on proprietary software and survey the first crop of open-source projects of comparable architecture and functionality. The main contribution of this paper is to show that openness is differentiated, and to offer scientific documentation of degrees of openness in this fast-moving field. We evaluate projects in terms of openness of code, training data, model weights, RLHF data, licensing, scientific documentation, and access methods. We find that while there is a fast-growing list of projects billing themselves as 'open source', many inherit undocumented data of dubious legality, few share the all-important instruction-tuning (a key site where human labour is involved), and careful scientific documentation is exceedingly rare. Degrees of openness are relevant to fairness and accountability at all points, from data collection and curation to model architecture, and from training and fine-tuning to release and deployment.

Abstract

Speech recognition systems are a key intermediary in voice-driven human-computer interaction. Although speech recognition works well for pristine monologic audio, real-life use cases in open-ended interactive settings still present many challenges. We argue that timing is mission-critical for dialogue systems, and evaluate 5 major commercial ASR systems for their conversational and multilingual support. We find that word error rates for natural conversational data in 6 languages remain abysmal, and that overlap remains a key challenge (study 1). This impacts especially the recognition of conversational words (study 2), and in turn has dire consequences for downstream intent recognition (study 3). Our findings help to evaluate the current state of conversational ASR, contribute towards multidimensional error analysis and evaluation, and identify phenomena that need most attention on the way to build robust interactive speech technologies.

Abstract

Iconicity in language is receiving increased attention from many fields, but our understanding of iconicity is only as good as the measures we use to quantify it. We collected iconicity measures for 304 Japanese words from English-speaking participants, using rating and guessing tasks. The words included ideophones (structurally marked depictive words) along with regular lexical items from similar semantic domains (e.g., fuwafuwa ‘fluffy’, jawarakai ‘soft’). The two measures correlated, speaking to their validity. However, ideophones received consistently higher iconicity ratings than other items, even when guessed at the same accuracies, suggesting the rating task is more sensitive to cues like structural markedness that frame words as iconic. These cues did not always guide participants to the meanings of ideophones in the guessing task, but they did make them more confident in their guesses, even when they were wrong. Consistently poor guessing results reflect the role different experiences play in shaping construals of iconicity. Using multiple measures in tandem allows us to explore the interplay between iconicity and these external factors. To facilitate this, we introduce a reproducible workflow for creating rating and guessing tasks from standardised wordlists, while also making improvements to the robustness, sensitivity and discriminability of previous approaches.

Abstract

What is language and who or what can be said to have it? In this essay we consider this question in the context of interactions with non-humans, specifically: animals and computers. While perhaps an odd pairing at first glance, here we argue that these domains can offer contrasting perspectives through which we can explore and reimagine language. The interactions between humans and animals, as well as between humans and computers, reveal both the essence and the boundaries of language: from examining the role of sequence and contingency in human-animal interaction, to unravelling the challenges of natural interactions with “smart” speakers and language models. By bringing together disparate fields around foundational questions, we push the boundaries of linguistic inquiry and uncover new insights into what language is and how it functions in diverse non-humanexclusive contexts.

Abstract

Prosociality and cooperation are key to what makes us human. But different cultural norms can shape our evolved capacities for interaction, leading to differences in social relations. How people share resources has been found to vary across cultures, particularly when stakes are high and when interactions are anonymous. Here we examine prosocial behavior among familiars (both kin and non-kin) in eight cultures on five continents, using video recordings of spontaneous requests for immediate, low-cost assistance (e.g., to pass a utensil). We find that, at the smallest scale of human interaction, prosocial behavior follows cross-culturally shared principles: requests for assistance are very frequent and mostly successful; and when people decline to give help, they normally give a reason. Although there are differences in the rates at which such requests are ignored, or require verbal acceptance, cultural variation is limited, pointing to a common foundation for everyday cooperation around the world.

Abstract

Iconicity, or the resemblance between form and meaning, is often ascribed to a special status and contrasted with default assumptions of arbitrariness in spoken language. But does iconicity in spoken language have a special status when it comes to learnability? A simple way to gauge learnability is to see how well something is retrieved from memory. We can further contrast this with guessability, to see (1) whether the ease of guessing the meanings of ideophones outperforms the rate at which they are remembered; and (2) how willing participants’ are to reassess what they were taught in a prior task—a novel contribution of this study. We replicate prior guessing and memory tasks using ideophones and adjectives from Japanese, Korean, and Igbo. Our results show that although native Cantonese speakers guessed ideophone meanings above chance level, they memorized both ideophones and adjectives with comparable accuracy. However, response time data show that participants took significantly longer to respond correctly to adjective–meaning pairs—indicating a discrepancy in a cognitive effort that favored the recognition of ideophones. In a follow-up reassessment task, participants who were taught foil translations were more likely to choose the true translations for ideophones rather than adjectives. By comparing the findings from our guessing and memory tasks, we conclude that iconicity is more accessible if a task requires participants to actively seek out sound-meaning associations.

Abstract

In individuals with synaesthesia specific sensory stimulation leads to unusual concurrent perceptions in the same or a different modality. Recent studies have demonstrated a high co-occurrence between synaesthesia and autism spectrum disorder (ASD), a condition also characterized by altered perception. A potentially shared characteristic of synaesthesia and ASD is a bias towards local (detail-focussed) perception. We investigated whether a bias towards local perception is indeed shared between synaesthesia and ASD. In a neurotypical population, we studied the relation between the degree of autistic traits (measured by the AQ) and the degree of grapheme-colour synaesthesia (measured by a consistency task), as well as whether both are related to a local bias in tasks assessing local/global visual perception. A positive correlation between total AQ scores and the degree of synaesthesia was found. Our study extends previous studies that found a high ASD-synaesthesia co-occurrence in clinical populations. Consistent with the hypothesized local perceptual bias in ASD, scores on the AQ-attention to detail subscale were related to increased performance on an Embedded Figures Task (EFT), and we found evidence for a relation to reduced susceptibility to visual illusions. We found no relation between autistic traits and local visual perception in a motion coherence task (MCT). Also, no relation between synaesthesia and local visual perception was found, although a reduced susceptibility to visual illusions resembled the results obtained for AQ-atttention to detail subscale. A suggested explanation for the absence of a relationship between the degree of synaesthesia and a local bias is that a possible local bias might be more pronounced in supra-threshold synaesthetes (compared to neurotypicals).

Abstract

While speculations on form–meaning resemblances in language go back millennia, the experimental study of iconicity is only about a century old. Here we take stock of experimental work on iconicity and present a double special issue with a diverse set of new contributions. We contextualise the work by introducing a typology of approaches to iconicity in language. Some approaches construe iconicity as a discrete property that is either present or absent; others treat it as involving semiotic relationships that come in kinds; and yet others see it as a gradient substance that comes in degrees. We show the benefits and limitations that come with each of these construals and stress the importance of developing accounts that can fluently switch between them. With operationalisations of iconicity that are well defined yet flexible enough to deal with differences in tasks, modalities, and levels of analysis, experimental research on iconicity is well equipped to contribute to a comprehensive science of language.

Abstract

Resource-rational approaches offer much promise for understanding human cognition, especially if they can reach beyond the confines of individual minds. Language allows people to transcend individual resource limitations by augmenting computation and enabling distributed cognition. Interactive language use, an environment where social rational agents routinely deal with resource constraints together, offers a natural laboratory to test resource-rationality in the wild.

Abstract

When people talk, they recruit a wide range of expressive devices for interactional work, from sighs, sniffs, clicks, and whistles to other conduct that borders on the linguistic. These resources represent some of the more elusive yet no less powerful aspects of the interactional machinery as they are used in the management of turn and sequence and the marking of stance and affect. Phenomena long assumed to be beyond the purview of linguistic inquiry emerge as systematically deployed practices whose ambiguous degree of control and convention allows participants to carry out subtle interactional work without committing to specific words. While these resources have been characterised as non-lexical, non-verbal, or non-conventional, I propose they are unified in their liminality: they work well precisely because they equivocate between sound and speech. The empirical study of liminal signs shows the promise of sequential analysis for building a science of language on interactional foundations.

Abstract

Aktuelle Vorstellungen von Gehirn-zu-Gehirn-Schnittstellen versprechen, die Sprache zu umgehen. Aber wenn wir sie verfeinern, um ihr kollaboratives Potenzial voll auszuschöpfen, sehen wir Sprache — oder zumindest ein sprachähnliches Infrastruktur für Kommunika­tion und Koordination — durch die Hintertür wieder hereinkommen. Es wäre nicht das erste Mal, dass sich die Sprache neu erfindet.

Current conceptions of brain-to-brain interfaces attempt to bypass language. But when we refine them to more fully realise their collaborative potential we find language —or at least a language-like infrastructure for communication and coordination— slipping through the back door. It wouldn't be the first time that language reinvented itself.

Abstract

Doing things for and with others is one of the foundations of human social life. This chapter studies a systematic collection of 207 requests for assistance and collaboration from a video corpus of everyday conversations in Siwu, a Kwa language of Ghana. A range of social action formats and semiotic resources reveals how language is adapted to the interactional challenges posed by recruiting assistance. While many of the formats bear a language-specific signature, their sequential and interactional properties show important commonalities across languages. Two tentative findings are put forward for further cross-linguistic examination: a “rule of three” that may play a role in the organisation of successive response pursuits, and a striking commonality in animal-oriented recruitments across languages that may be explained by convergent cultural evolution. The Siwu recruitment system emerges as one instance of a sophisticated machinery for organising collaborative action that transcends language and culture.

Abstract

Words like ‘waddle’, ‘flop’ and ‘zigzag’ combine playful connotations with iconic form-meaning resemblances. Here we propose that structural markedness may be a common factor underlying perceptions of playfulness and iconicity. Using collected and estimated lexical ratings covering a total of over 70,000 English words, we assess the robustness of this assocation. We identify cues of phonotactic complexity that covary with funniness and iconicity ratings and that, we propose, serve as metacommunicative signals to draw attention to words as playful and performative. To assess the generalisability of the findings we develop a method to estimate lexical ratings from distributional semantics and apply it to a dataset 20 times the size of the original set of human ratings. The method can be used more generally to extend coverage of lexical ratings. We find that it reliably reproduces correlations between funniness and iconicity as well as cues of structural markedness, though it also amplifies biases present in the human ratings. Our study shows that the playful and the poetic are part of the very texture of the lexicon.

Abstract

This article makes the case for the universality of the sequence organization observable in informal human conversational interaction. Using the descriptive schema developed by Schegloff (2007), we examine the major patterns of action-sequencing in a dozen nearly all unrelated languages. What we find is that these patterns are instantiated in very similar ways for the most part right down to the types of different action sequences. There are also some notably different cultural exploitations of the patterns, but the patterns themselves look strongly universal. Recent work in gestural communication in the great apes suggests that sequence organization may have been a crucial route into the development of language. Taken together with the fundamental role of this organization in language acquisition, sequential behavior of this kind seems to have both phylogenetic and ontogenetic priority, which probably puts substantial functional pressure on language form.

Abstract

When people are engaged in social interaction, they can repeat aspects of each other’s communicative behavior, such as words or gestures. This kind of behavioral alignment has been studied across a wide range of disciplines and has been accounted for by diverging theories. In this paper, we review various operationalizations of lexical and gestural alignment. We reveal that scholars have fundamentally different takes on when and how behavior is considered to be aligned, which makes it difficult to compare findings and draw uniform conclusions. Furthermore, we show that scholars tend to focus on one particular dimension of alignment (traditionally, whether two instances of behavior overlap in form), while other dimensions remain understudied. This hampers theory testing and building, which requires a well‐defined account of the factors that are central to or might enhance alignment. To capture the complex nature of alignment, we identify five key dimensions to formalize the relationship between any pair of behavior: time, sequence, meaning, form, and modality. We show how assumptions regarding the underlying mechanism of alignment (placed along the continuum of priming vs. grounding) pattern together with operationalizations in terms of the five dimensions. This integrative framework can help researchers in the field of alignment and related phenomena (including behavior matching, mimicry, entrainment, and accommodation) to formulate their hypotheses and operationalizations in a more transparent and systematic manner. The framework also enables us to discover unexplored research avenues and derive new hypotheses regarding alignment.

Abstract

How can people communicate successfully while keeping resource costs low in the face of ambiguity? We present a principled theoretical analysis comparing two strategies for disambiguation in communication: (i) pragmatic reasoning, where communicators reason about each other, and (ii) other-initiated repair, where communicators signal and resolve trouble interactively. Using agent-based simulations and computational complexity analyses, we compare the efficiency of these strategies in terms of communicative success, computation cost and interaction cost. We show that agents with a simple repair mechanism can increase efficiency, compared to pragmatic agents, by reducing their computational burden at the cost of longer interactions. We also find that efficiency is highly contingent on the mechanism, highlighting the importance of explicit formalisation and computational rigour.

Abstract

Reasons play an important role in social interaction. We study reasons-giving in the context of request sequences in Russian. By contrasting request sequences with and without reasons, we are able to shed light on the interactional work people do when they provide reasons or ask for them. In a systematic collection of request sequences in everyday conversation (N = 158), we find reasons in a variety of sequential positions, showing the various points at which participants may orient to the need for a reason. Reasons may be left implicit (as in many minimal requests that are readily complied with), or they can be made explicit. Participants may make reasons explicit either as part of the initial formulation of a request or in an interactionally contingent way. Across sequential positions, we show that reasons for requests recurrently deal with three possible issues: (1) providing information when a request is underspecified, (2) managing relationships between the requester and requestee and (3) explicating ancillary actions implemented by a request. By spelling out information normally left to presuppositions and implicatures, reasons make requests more understandable and help participants to navigate the social landscape of asking assistance from others.

Abstract

We provide an annotated coding scheme for other-initiated repair, along with guidelines for building collections and aggregating cases based on interactionally relevant similarities and differences. The questions and categories of the scheme are grounded in inductive observations of conversational data and connected to a rich body of work on other-initiated repair in conversation analysis. The scheme is developed and tested in a 12-language comparative project and can serve as a stepping stone for future work on other-initiated repair and the systematic comparative study of conversational structures.

Abstract

Sound symbolism is a phenomenon with broad relevance to the study of language and mind, but there has been a disconnect between its investigations in linguistics and psychology. This study tests the sound-symbolic potential of ideophones—words described as iconic—in an experimental task that improves over prior work in terms of ecological validity and experimental control. We presented 203 ideophones from five languages to eighty-two Dutch listeners in a binary-choice task, in four versions: original recording, full diphone resynthesis, segments-only resynthesis, and prosody-only resynthesis. Listeners guessed the meaning of all four versions above chance, confirming the iconicity of ideophones and showing the viability of speech synthesis as a way of controlling for segmental and suprasegmental properties in experimental studies of sound symbolism. The success rate was more modest than prior studies using pseudowords like bouba/kiki, implying that assumptions based on such words cannot simply be transferred to natural languages. Prosody and segments together drive the effect: neither alone is sufficient, showing that segments and prosody work together as cues supporting iconic interpretations. The findings cast doubt on attempts to ascribe iconic meanings to segments alone and support a view of ideophones as words that combine arbitrariness and iconicity.We discuss the implications for theory and methods in the empirical study of sound symbolism and iconicity.

Abstract

Sound symbolism is increasingly understood as involving iconicity, or perceptual analogies and cross-modal correspondences between form and meaning, but the search for its functional and neural correlates is ongoing. Here we study how people learn sound-symbolic words, using behavioural, electrophysiological and individual difference measures. Dutch participants learned Japanese ideophones —lexical sound-symbolic words— with a translation of either the real meaning (in which form and meaning show cross-modal correspondences) or the opposite meaning (in which form and meaning show cross-modal clashes). Participants were significantly better at identifying the words they learned in the real condition, correctly remembering the real word pairing 86.7% of the time, but the opposite word pairing only 71.3% of the time. Analysing event-related potentials (ERPs) during the test round showed that ideophones in the real condition elicited a greater P3 component and late positive complex than ideophones in the opposite condition. In a subsequent forced choice task, participants were asked to guess the real translation from two alternatives. They did this with 73.0% accuracy, well above chance level even for words they had encountered in the opposite condition, showing that people are generally sensitive to the sound-symbolic cues in ideophones. Individual difference measures showed that the ERP effect in the test round of the learning task was greater for participants who were more sensitive to sound symbolism in the forced choice task. The main driver of the difference was a lower amplitude of the P3 component in response to ideophones in the opposite condition, suggesting that people who are more sensitive to sound symbolism may have more difficulty to suppress conflicting cross-modal information. The findings provide new evidence that cross-modal correspondences between sound and meaning facilitate word learning, while cross-modal clashes make word learning harder, especially for people who are more sensitive to sound symbolism.

Abstract

The existence of sound-symbolism (or a non-arbitrary link between form and meaning) is well-attested. However, sound-symbolism has mostly been investigated with nonwords in forced choice tasks, neither of which are representative of natural language. This study uses ideophones, which are naturally occurring sound-symbolic words that depict sensory information, to investigate how sensitive Dutch speakers are to sound-symbolism in Japanese in a learning task. Participants were taught 2 sets of Japanese ideophones; 1 set with the ideophones’ real meanings in Dutch, the other set with their opposite meanings. In Experiment 1, participants learned the ideophones and their real meanings much better than the ideophones with their opposite meanings. Moreover, despite the learning rounds, participants were still able to guess the real meanings of the ideophones in a 2-alternative forced-choice test after they were informed of the manipulation. This shows that natural language sound-symbolism is robust beyond 2-alternative forced-choice paradigms and affects broader language processes such as word learning. In Experiment 2, participants learned regular Japanese adjectives with the same manipulation, and there was no difference between real and opposite conditions. This shows that natural language sound-symbolism is especially strong in ideophones, and that people learn words better when form and meaning match. The highlights of this study are as follows: (a) Dutch speakers learn real meanings of Japanese ideophones better than opposite meanings, (b) Dutch speakers accurately guess meanings of Japanese ideophones, (c) this sensitivity happens despite learning some opposite pairings, (d) no such learning effect exists for regular Japanese adjectives, and (e) this shows the importance of sound-symbolism in scaffolding language learning

Abstract

Studies of sound symbolism have shown that people can associate sound and meaning in consistent ways when presented with maximally contrastive stimulus pairs of nonwords such as bouba/kiki (rounded/sharp) or mil/mal (small/big). Recent work has shown the effect extends to antonymic words from natural languages and has proposed a role for shared cross-modal correspondences in biasing form-to-meaning associations. An important open question is how the associations work, and particularly what the role is of sound-symbolic matches versus mismatches. We report on a learning task designed to distinguish between three existing theories by using a spectrum of sound-symbolically matching, mismatching, and neutral (neither matching nor mismatching) stimuli. Synthesized stimuli allow us to control for prosody, and the inclusion of a neutral condition allows a direct test of competing accounts. We find evidence for a sound-symbolic match boost, but not for a mismatch difficulty compared to the neutral condition.

Abstract

Numerous studies demonstrate people associate colors with letters and numbers in systematic ways. But most of these studies rely on speakers of English, or closely related languages. This makes it difficult to know how generalizable these findings are, or what factors might underlie these associations. We investigated letter–color and number–color associations in Arabic speakers, who have a different writing system and unusual word structure compared to Standard Average European languages. We also aimed to identify grapheme–color synaesthetes (people who have conscious color experiences with letters and numbers). Participants associated colors with 28 basic Arabic letters and ten digits by typing color names that best fit each grapheme. We found language-specific principles determining grapheme–color associations. For example, the word formation process in Arabic was relevant for color associations. In addition, psycholinguistic variables, such as letter frequency and the intrinsic order of graphemes influenced associations. Contrary to previous studies we found no evidence for sounds playing a role in letter–color associations for Arabic, and only a very limited role for shape influencing color associations. These findings highlight the importance of linguistic and psycholinguistic features in cross-modal correspondences, and illustrate why it is important to play close attention to each language on its own terms in order to disentangle language-specific from universal effects