No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Partner‐Specific Adaptation in Dialog
Abstract
No one denies that people adapt what they say and how they interpret what is said to them, depending on their interactive partners. What is controversial is when and how they do so. Several psycholinguistics research programs have found what appear to be failures to adapt to partners in the early moments of processing and have used this evidence to argue for modularity in the language processing architecture, claiming that the system cannot take into account a partner’s distinct needs or knowledge early in processing. We review the evidence for both early and delayed partner-specific adaptations, and we identify some challenges and difficulties with interpreting this evidence. We then discuss new analyses from a previously published referential communication experiment (Metzing & Brennan, 2003) demonstrating that partner-specific effects need not occur late in processing. In contrast to Pickering and Garrod (2004) and Keysar, Barr, and Horton (1998b), we conclude that there is no good evidence that early processing has to be be “egocentric,”“dumb,” or encapsulated from social knowledge or common ground, but that under some circumstances, such as when one partner has made an attribution about another’s knowledge or needs, processing can be nimble enough to adapt quite early to a perspective different from one’s own.
... To circumvent some of these challenges, shared properties of explananda are interesting. The content of an explanation is influenced by the explanandum (the subject of explanation), more precisely the EX's mental representation of that explanandum, and the knowledge needs of the Explainee (EE) regarding that explanandum [3,10]. Whenever the explanandum is a technological artifact, one can refer to its dual nature, following a techno-philosophical theory [22,46]. ...
... The EE's instruction was, "[i]n a moment, a person will enter the room and explain a board game to you. Please participate actively in the explanation" before the EX entered the room 3 . ...
... Original: "In dem nächsten Raum erklären Sie bitte dem Gegenüber das Spiel so gut, dass Ihr Gegenüber eine Chance hätte, das Spiel zu gewinnen".3 Original: "Im nächsten Raum ist eine Person, die Ihnen ein Spiel erklärt, bitte nehmen Sie aktiv an der Erklärung teil". ...
In XAI it is important to consider that, in contrast to explanations for professional audiences, one cannot assume common expertise when explaining for laypeople. But such explanations between humans vary greatly, making it difficult to research commonalities across explanations. We used the dual nature theory, a techno-philosophical approach, to cope with these challenges. According to it, onan explain, for example, an XAI’s decision by addressing its dual nature: by focusing on the Architecture (e.g., the logic of its algorithms) or the Relevance (e.g., the severity of a decision, the implications of a recommendation). We investigated 20 explanations of games using the theory as an analytical framework. We elaborate how we used the theory to quickly structure and compare explanations of technological artifacts. We supplement results from analyzing the explanation contents with results from a video recall to explore how Explainers (EX) justified their explanation. We found that EX were focusing on the physical aspects of the game first (Architecture) and only later on aspects of the Relevance. Reasoning in the video recalls indicated that EX regarded the focus on the Architecture as important for structuring the explanation initially by explaining the basic components before focusing on more complex, intangible aspects. EX justified shifting between addressing the two sides by explanation goals, emerging misunderstandings, and the knowledge needs of the explainee. We discovered several commonalities that inspire future research questions which, if further generalizable, provide first ideas for the construction of synthetic explanations.
... The latter are a set of rules and processes that enable communication between two agents or systems, whether technological, robotic or human (Bochmann & Sunshine, 1980). For example, if we consider a conversation between people, the communication protocol is made up of three elements: what is said, how it is said, and the characteristics of who it is said to (Brennan & Hanna, 2009). If this protocol is transferred to an HRI, what is said is pre-programmed in the so-called 'script', and how it is said is determined by the text-to-speech software, which can also be combined with non-verbal language signals (Chidambaram et al., 2012;Forgas-Coll et al., 2022a). ...
... If this protocol is transferred to an HRI, what is said is pre-programmed in the so-called 'script', and how it is said is determined by the text-to-speech software, which can also be combined with non-verbal language signals (Chidambaram et al., 2012;Forgas-Coll et al., 2022a). However, the process of the third elementknowing the characteristics of the audienceis more complex, and requires clear expectations about their level of knowledge of the topic or their ability to understand what is being said (Brennan & Hanna, 2009). ...
... For instance, chatbots are not expected to reach credible levels of human intelligence before 2029 (Shridhar, 2017). However, one way to improve the perceived performance of social robots is for designers to make greater consideration of users' own skills in relation to the corresponding task, and also their ability and predisposition with regard to HRIs, this being the third element of the communication protocol (Brennan & Hanna, 2009). Evidence has been found in the literature that the context, task and target audience condition the effectiveness of communication protocols. ...
The ability to install social intelligence protocols in robots in order for them to exhibit conversational skills has made them ideal tools for delivering services with a high cognitive and low emotional load. Little is known about how this capability influences the customer experience and the intention to continue receiving these services. Experiences were assessed in a study simulating customer-facing service delivery, and the constructs of the technology readiness index and stated gender were analysed as possible moderators in a quasi-experiment. Hedonic quality was the most relevant factor explaining attitude, and attitude explained intention to use as well as social influence. As for the constructs of technological readiness and gender, optimism and innovativeness seem to be the most likely candidates for moderating the other variables. The most optimistic and the most innovative route would be for the main actors to continue adapting to social robot technology in the future.
... At first glance, a simple handshake does not have much in common with an expertly synchronized swimming performance nor with an improvised dinner party. However, what co-actors in all types of joint actions rely on is the fact that they share something: Individuals "cannot even begin to coordinate […] without assuming a vast amount of shared information or common ground" (Clark & Brennan, 1991, p. 222; cf. also Brennan & Hanna, 2009;Clark, 1996, p. 199;Lewis, 1969;Schelling, 1960;Stalnaker, 2002). Common ground is crucial to get a joint action started; at the same time, new common ground builds up between co-actors with every joint action they perform. ...
... Thus, individuals engaged in joint action will act upon what they think is common ground in the specific interaction and they will therefore expect that their behavior is comprehensible for their interaction partners. This is most obviously the case in conversations where speakers adjust to the expected shared background with their addressees, for example by simplifying word choices when interacting with children (Brennan & Hanna, 2009;Clark & Krych, 2004;Horton, 2007;Horton & Keysar, 1996;Keysar, Barr, Balin, & Brauner, 2000;Lockridge & Brennan, 2002). It is, however, possible that an individual's belief about what is common ground in an interaction is actually false and that they will thus not be understood (or misunderstood) by their interaction partnerit is only when both partners' beliefs about their common ground coincide that the interaction will be successful. ...
... Such examples vividly demonstrate how senders' and receivers' expectations need to be aligned for communication to succeed. Importantly, senders need to be aware of the receivers' background (that might differ from their own) and need to carefully monitor the receivers' behavior in order to adjust their communicative signals if necessary, just as in conventional linguistic communication (Brennan & Hanna, 2009;Brennan, Kuhlen, & Charoy, 2018;Clark & Kruch, 2004;Lockridge & Brennan, 2002). In the present study, senders designed their communicative signals based on the (implicit) assumption that crossmodal correspondences would provide a background shared by receivers. ...
When performing joint actions, people rely on common ground-shared information that provides the required basis for mutual understanding. Common ground can be based on people's interaction history or on knowledge and expectations people share, e.g., because they belong to the same culture or social class. Here, we suggest that people rely on yet another form of common ground, one that originates in their similarities in multisensory processing. Specifically, we focus on 'crossmodal correspondences' - nonarbitrary associations that people make between stimulus features in different sensory modalities, e.g., between stimuli in the auditory and the visual modality such as high-pitched sounds and small objects. Going beyond previous research that focused on investigating crossmodal correspondences in individuals, we propose that people can use these correspondences for communicating and coordinating with others. Initial support for our proposal comes from a communication game played in a public space (an art gallery) by pairs of visitors. We observed that pairs created nonverbal communication systems by spontaneously relying on 'crossmodal common ground'. Based on these results, we conclude that crossmodal correspondences not only occur within individuals but that they can also be actively used in joint action to facilitate the coordination between individuals.
... Similarly, investigating the clarity of words that have been mentioned in a corpus twice, also (Bard et al., 2000) find that speakers rely more on the fact that the word was familiar to themselves rather than to the addressee. In contrast, Brennan and Hanna (2009) argue that Bard et al. (2000) failed to provide a comparison, i.e. a control group to whom speakers also told the story a second time but who had not heard the story before; compared with speech directed to different addressees, there is indeed a clear difference between second mention to the same versus to a different addressee. Galati and Brennan (2009) investigate storytelling to same and to different addressees and find significant support for partner-specific attenuation, i.e. adaptation to the respective listener's information needs. ...
... Several authors attribute the many null findings with respect to addressee orientation to methodological problems in the analysis: Many studies rely on imagined addressees or make use of confederates, who may subconsciously influence the participants' linguistic productions, or on monological, noninteractive contexts. As several authors, for instance, Schober and Brennan (2003), Brown-Schmidt (2009) and Brennan and Hanna (2009), criticize, confederates may have influenced subconsciously the production process in many of the experiments reported on above. Furthermore, the results from Brown and Dell (1987) or Horton and Keysar (1996) neglect the interlocutors' ability to interact, which may have a great impact on the representation of addressees' epistemic and deontic states, as Brown-Schmidt (2009) argues; she compares partner-specific effects in interactive and non-interactive settings and finds that partner-specific interpretation is likely in interactive dialog settings whereas it often cannot be found in non-interactive settings. ...
... That is, while we can manipulate a robot's appearance, behavior and linguistic output, we cannot manipulate human communication partners to the same extent, at least it cannot be guaranteed that a confederate, for instance, will not provide cues that cast doubt on the validity of the results (cf. Brennan and Hanna, 2009). Thus, using humanrobot interaction to investigate variables of addressee orientation seems to be useful. ...
... Linguistic synchrony is the degree to which two or more interlocutors reciprocally share linguistic properties, such as lexical choices in dialogue (Branigan et al., 2000;Brennan & Hanna, 2009;Pickering & Garrod, 2004). Linguistic synchrony is often referred to as interactive alignment or shared language use (Bayram & Ta, 2020). ...
... Linguistic synchrony, also referred to as linguistic alignment , linguistic entrainment (Hirschberg et al., 2008;Rahimi et al., 2017), or shared language use (Bayram & Ta, 2020), is the degree to which two or more interlocutors reciprocally share linguistic properties such as lexical choices in a dialogue (Branigan et al., 2000;Brennan & Hanna, 2009;Pickering & Garrod, 2004). Linguistic synchrony is known to be correlated with better collaboration , common knowledge building (Bayram & Ta, 2020), task success (Hirschberg et al., 2008) and inter-subjective meaning in context (Ickes, 2002). ...
Previous literature has associated math literacy with linguistic factors such as verbal ability and phonological skills. However, few studies have investigated linguistic synchrony , shown in mathematical discussions. This study modelled math literacy and examined the relationship of math literacy with linguistic synchrony between students and facilitators. We retrieved data from 20,776 online mathematical discussion threads at a secondary school level. First, we assessed students' math literacy based on their discussions and classified them into high‐ and low‐math literacy groups. Then, we conducted Cross‐Recurrence Quantification Analysis (CRQA) to calculate linguistic synchrony within each thread. The result implies that students with high math literacy are more likely to share common words (eg, mathematical terms) with facilitators. At the same time, they would paraphrase the facilitators' words rather than blindly mimic them as the exact sentences or phrases. On the other hand, students with low math literacy tend to use overlapping words with facilitators less frequently and are more likely to repeat the exact same phrases from the facilitators. The findings provide an empirical data analysis and insights into mathematical discussions and linguistic synchrony. In addition, this paper implies the directions to improve online mathematical discussions and foster math literacy.
Practitioner notes
What is already known about this topic Mathematical discussions are known to be an effective way to promote math literacy.
Math literacy and linguistic skills have a strong link.
Linguistic synchrony is related to better collaboration and common knowledge building.
What this paper adds Reveals the relationship between math literacy and linguistic synchrony and deepens the understanding of digital communication in online learning environments.
Provides empirical analysis of natural language data in group discussions using CRQA.
Conceptualizes linguistic synchrony with three sub‐concepts: linguistic concurrence, predictability, and complexity.
Implications for practice and/or policy Educators and practitioners could utilize the automatic formative assessment of math literacy based on the student's language use in mathematical discussions.
Educational technology researchers and designers could include CRQA indices and recurrence plots in the dashboard design to provide information to support teachers and learners.
Teachers would be able to provide real‐time interventions to promote effective mathematical communication and foster math literacy throughout mathematical discussions.
... Following this, we provide a general discussion and close with recommendations of avenues for future research. misunderstandings or speech errors (Brennan & Hanna, 2009). In each of these scenarios, linguistic coordination is linked to cognitive coordination (see also Hiver & Al-Hoorie, 2020). ...
... Conflict resolution was identified as another circumstance under which co-adaptation occurred in the email exchange (see also Brennan & Hanna, 2009). As shown in Figure 4, in Debra's first email (1a), she expressed her confusion and disappointment over not receiving an email from Alicia, as it was expected that Alicia would initiate the communication. ...
In the field of applied linguistics, Diane Larsen-Freeman is widely recognized, among her numerous contributions, as the originator of Complex Dynamic Systems Theory (CDST). Over the years since her seminal publication (Larsen-Freeman, 1997), CDST has evolved into a meta-theory that guides the field’s thinking on a variety of developmental phenomena, not least the study of second language acquisition. As its theoretical potency rises, so does the need to achieve a tangible understanding of CDST’s core constructs. In this paper, we empirically examine one such construct, co-adaptation, in the context of asynchronous dyadic interaction. A data corpus of 39 emails sent between two college students in China and the United States over the course of seven weeks is analyzed using NVivo 12. Co-adaptation is observed at the pragmatic, discourse, and linguistic levels. True to the spirit of CDST, we describe, and discuss the nuances of, our findings. Considering the novelty of our approach to data analysis, we end by reflecting on the limitations specific to our study, as well as worthwhile directions for future pursuits.
... To circumvent some of these issues, shared properties of explananda are interesting. The content of an explanation is influenced by the explanandum (the subject of explanation), more precisely the EX's mental representation of that explanandum, and the knowledge needs of the Explainee (EE) regarding that explanandum [9,3]. Whenever the explanandum is a technological artifact, one can refer to its dual nature, following a techno-philosophical theory [21,44]. ...
... The EE's instruction was, "[i]n a moment, a person will enter the room and explain a board game to you. Please participate actively in the explanation" before the EX entered the room 3 . ...
In XAI it is important to consider that, in contrast to explanations for professional audiences, one cannot assume common expertise when explaining for laypeople. But such explanations between humans vary greatly, making it difficult to research commonalities across explanations. We used the dual nature theory, a techno-philosophical approach, to cope with these challenges. According to it, one can explain, for example, an XAI's decision by addressing its dual nature: by focusing on the Architecture (e.g., the logic of its algorithms) or the Relevance (e.g., the severity of a decision, the implications of a recommendation). We investigated 20 game explanations using the theory as an analytical framework. We elaborate how we used the theory to quickly structure and compare explanations of technological artifacts. We supplemented results from analyzing the explanation contents with results from a video recall to explore how explainers justified their explanation. We found that explainers were focusing on the physical aspects of the game first (Architecture) and only later on aspects of the Relevance. Reasoning in the video recalls indicated that EX regarded the focus on the Architecture as important for structuring the explanation initially by explaining the basic components before focusing on more complex, intangible aspects. Shifting between addressing the two sides was justified by explanation goals, emerging misunderstandings, and the knowledge needs of the explainee. We discovered several commonalities that inspire future research questions which, if further generalizable, provide first ideas for the construction of synthetic explanations.
... However, it is not always clear whether the referential choices that speakers make are listener-driven or simply context-driven, as both the listener and the speaker usually have access to the same discourse context (Hendriks et al., 2014: 392). There is evidence that referential choices and their prosodic marking are, at least to some degree, listener-driven (see, e.g., Hendriks et al., 2014;Galati and Brennan, 2010; for a review see Brennan and Hanna, 2009). Some have argued for models in which the early, rapid processing is susceptible to egocentric speaker-only knowledge, and more effortful listenerdriven adjustments emerge relatively later (e.g. ...
... However, it is not always clear whether the referential choices that speakers make are listener-driven or simply context-driven, as both the listener and the speaker usually have access to the same discourse context (Hendriks et al., 2014: 392). There is evidence that referential choices and their prosodic marking are, at least to some degree, listener-driven (see, e.g., Hendriks et al., 2014;Galati and Brennan, 2010, for a review see Brennan and Hanna (2009)). The present study supports this assumption in two ways. ...
The aim of this thesis is to contribute to the current knowledge of prosody competence in high-functioning individuals with Autism Spectrum Disorders (ASD) by investigating the perception and production of referential givenness in this population. Previous studies have shown that the encoding and decoding of information structure by means of prosody poses an area that is particularly affected in adults with ASD. In a perception experiment, participants had to make judgements as to how far an item or person referred to by a target word sounds as if it is known or new. Results reveal that participants from the ASD group made significantly less use of prosody than participants from the control group did. In a production experiment, the ability of adults with ASD to encode referential givenness was investigated in a cooperative story-telling task. The findings for both the prosodic marking and the choice of referring expressions indicate that most of the speakers with ASD did not attenuate given information to the same extent as control speakers did. Taken together, the two experiments presented in this thesis provide further evidence for the assumption that pragmatic prosody represents an area of particular difficulty for individuals with ASD. Observations from both experiments have also confirmed that individuals with ASD tend to employ compensation mechanisms both in structured tasks and in every-day social encounters. While at first view, their use of prosody might not always appear strikingly deviant, a profound and detailed analysis might reveal subtle differences that, in sum, can lead to the impression of a speaker being less involved in conversation. The findings highlight the presence of prosodic deficits even in high-functioning adults with ASD and might help to better understand the difficulties encountered by people with ASD in speech-based communication and social encounters.
... In the native condition, he produces a metaphoric that represents the concept of rock-climbing (like holding the idea) whereas in the non-native condition, he illustrates the action of rock-climbing with an iconic. Thus, this example is a good illustration of accommodation phenomenon, especially of complementarity (Brennan & Hanna 2009) since the gesture produced with speech conveys more concrete meaning in the non-native condition, as if the speaker intended to help oral understanding. It is noteworthy that the accommodation is performed only through the gestural channel. ...
... The future teachers adjusted their gestures to meet the needs of their non-native interlocutors. These adjustments clearly illustrate complementary discourse (Brennan & Hanna, 2009). ...
This paper addresses the question of how speakers adapt their gestures according to their interlocutor’s proficiency level in the language of the interaction especially in the specific context of foreign language teaching. We know that speakers make changes in their speech when addressing a non-native speaker, called Foreigner Talk ( Ferguson, 1975 ) to make their speech more comprehensible. However, whether and how gestures are also modified along with speech has hardly been addressed in the literature. In this study, we examined the speech and gesture of future teachers of French in a word explanation task to see what types of adjustments they made when explaining a word to a native speaker and a non-native speaker. We had ten future teachers of French explain the same 12 words to a native and a non-native speaker of French and compared the explanations. We found that the future teachers produced significantly more gestures, significantly longer gestures in duration, significantly more illustrative (iconic and deictic) gestures, and significantly larger gestures when addressing a non-native interlocutor. These results show that native speakers make not only speech adjustments but also gesture adjustments in addressing non-native speakers.
... Here, goals can both represent external, shared goals that are, for example, provided by the company management in an organization, or more implicit individual goals that develop throughout the interaction. A collective use of function words-representing a synchronized style of speaking-indicates a common understanding and worldview (Brennan and Hanna 2009). In relation to Ms. Flach's team meetings this means that the team members coordinate their implicit communication in the form of function word use depending on its context or goal in order to achieve their respective goals. ...
... Taking a closer look at the temporal sequence of language style matching, it is still striking that with scores ≥ 0.89 language style matching is high over the course of this excerpt. Following our explanation of language style matching, this might be indicative of a shared worldview-in this short excerpt of the meeting, the team members linguistically relate to each other and seem to have a shared understanding of the conversational content (Brennan and Hanna 2009). ...
This report in the journal Gruppe. Interaktion. Organisation. Zeitschrift für Angewandte Organisationspsychologie aims at presenting how the analysis of implicit and explicit communication in organizational interaction can advance our insights into and implications for these interactions for research and science. Communication is a central process in modern organizations. Especially recurring forms of interaction in organizations (e.g., meetings or appraisal interviews) are of great importance for personal and organizational success. In these interactions, the communication between the interacting organizational members has a decisive impact on the interactions’ course and outcomes (e.g., satisfaction with the interaction, performance during the interaction). Therefore, the aim of this paper is to present two aspects of communication that are empirically shown to contribute to successful outcomes of organizational interactions. Based on a practical problem, we illustrate the analysis and implications of (1) implicit communication (that is, the use and coordination of unconsciously used function words such as pronouns, articles, or prepositions) and (2) explicit communication (that is, the overarching meaning of a statement). To further illustrate the practical relevance of both communication behaviors, we present empirical insights and their implications for practice. Taking a glance at the future, possible combinations of these communication behaviors, the resulting avenues for future research, and the importance of a strengthened cooperation between research and practice to gain more naturalistic insights into organizational communication dynamics are discussed.
... Still, there is ample evidence that speakers, at least in certain situations, adjust their referential choices to properties of the listener's perspective (e.g., Ahn & Brown-Schmidt, 2020;Brennan & Hanna, 2009;Hawkins et al., 2021;Hendriks et al., 2014;Loy et al., 2020;Tal et al., 2023;Vogels et al., 2020). In this paper, I have assumed that taking into account the predictability of referents is related to speakers taking the perspective of their addressee (cf. ...
Language comprehension involves continuously making predictions about what will be mentioned next. If speakers take these predictions into account, one would expect that they try to be extra clear (e.g., by saying “the girl with the big earrings”) when they are going to say something less predictable. Conversely, speakers do not need to be as clear when the listener already expects the thing that they are about to mention, and can therefore suffice with a pronoun such as she. Previous research testing this hypothesis has found mixed results, with some studies finding that the referent’s predictability in discourse affects pronoun use and others finding that it does not. One explanation might be that speakers are more likely to take predictability into account when there is a co-present addressee who is predicting the next referent. To test this possibility, I conducted a language production experiment in which participants produced spoken continuations of narrative fragments. The fragments were accompanied by pictures that made clear how the story continued. Half of the participants performed the task without anyone else being present, while the other half told the stories to another person, who had to pick out the correct picture. Referent predictability was varied by manipulating the coherence relation in the narrative context. In addition, I calculated a surprisal score for each character in each narrative, as a more direct measure of its predictability. The results showed that with higher predictability, speakers were indeed more likely to use a pronoun than a definite NP to refer to the target character in their continuations. However, it did not matter whether the speaker was telling the stories to a co-present addressee or not. The results are discussed in light of accounts that distinguish between taking the perspective of a specific and that of a hypothetical listener.
... However, the linguistic adaptations can also be explained from the perspective of the Communication Accommodation Theory (Giles et al., 1991). This theoretical approach assumes that the speaker and receiver of a message relate to each other in the communication process and speakers consider the receivers' perspective, level of knowledge, and specific needs, with considerations usually based on stereotyped representations (Brennan & Hanna, 2009;Giles et al., 1991). Tellier et al. (2021) have argued that although this theory has been explicitly designed to cover verbal and nonverbal action, the body of studies has almost exclusively examined verbal communication adaptations, while there is a lack of research on nonverbal communication. ...
Teachers’ use of gestures in the classroom can support the language acquisition of learners in learning a second language ( Stam & Tellier, 2022 ). Depending on learners’ language skills, different dimensions of gestures (e.g., deictic, metaphorical) are considered to facilitate successful language comprehension. This study investigates which gestures teachers use in German as a second language (GSL) classrooms and to what extent teachers adapt their gestures to learners’ language proficiency. Teacher gestures in 10 video-recorded integration and preparation classes were analyzed. Two coders reliably identified 4143 gestures. Results show that GSL teachers predominantly used deictic gestures, metaphorical gestures, and feedback by head movements. Moreover, between-learner variability in teachers’ use of deictic and metaphorical gestures was explained by teacher-perceived German language proficiency of learners. These results suggest that teachers systematically adapt some dimensions of gestures in GSL classes, thus emphasizing the importance of studying nonverbal interactions for a better understanding of language acquisition processes.
... Empirical findings show a number of recurring behavioral trends in the reference game task. For example, descriptions are dramatically shortened across iterations (Krauss and Weinheimer, 1964), and the resulting labels are partner-specific (Wilkes-Gibbs and Clark, 1992;Brennan and Hanna, 2009). ...
... Interpersonal synchronization is thought to promote mutual comprehension Brennan & Hanna, 2009) and dyadic interactions in verbal tasks (Abney et al., 2014;Pickering & Garrod, 2004;Shockley et al., 2002), interaction quality (Reuzel et al., 2013) and an increased sense of "togetherness" (Ravreby et al., 2022). We outlined that some of the effects we observed could be driven by mutual influence (Nowak et al., 2020;Thibaut & Kelley, 1959) and understood by means of dyadic coupling (e.g., Abney et al., 2014;Konvalinka et al., 2011;Schloesser et al., 2019). ...
Objective
In social interactions, humans tend to naturally synchronize their body movements. We investigated interpersonal synchronization in conversations and examined its relationship with personality differences and post‐interaction appraisals.
Method
In a 15‐minute semi‐structured conversation, 56 previously‐unfamiliar dyads introduced themselves, followed by self‐disclosing and argumentative conversations. Their bodily movements were video‐recorded in a standardized room (112 young adults, aged 18–33, mean = 20.54, SD = 2.74; 58% Dutch, 31% German, 11% other). Interpersonal bodily synchronization was estimated as (a) synchronization strength using Windowed Lagged Cross‐Correlations and (b) Dynamic Organization (Determinism/Entropy/Laminarity/Mean Line) using Cross‐Recurrence Quantification Analysis. Bodily synchronization was associated with differences in Agreeableness and Extraversion (IPIP‐NEO‐120) and post‐conversational appraisals (affect/closeness/enjoyment) in mixed‐effect models.
Results
Agreeable participants exhibited higher complexity in bodily synchronization dynamics (higher Entropy) than disagreeable individuals, who also reported more negative affect afterward. Interpersonal synchronization was stronger among extroverts than among introverts and extroverts appraised conversations as more positive and enjoyable. Bodily synchronization strength and dynamic organization were related to the type of conversation (self‐disclosing/argumentative).
Conclusions
Interpersonal dynamics were intimately connected to differences in Agreeableness and Extraversion, varied across situations, and these parameters affected how pleasant, close, and enjoyable each conversation felt.
... Despite the aforementioned studies associating students' math literacy with a series of linguistic factors, there is no research analyzing the relationship between linguistic synchrony and math literacy. Linguistic synchrony, also referred to as interactive alignment [15] or shared language use [3], is the degree to which two or more interlocutors reciprocally share linguistic properties, such as lexical choices in dialogue [6,7,32]. Linguistic synchrony is known to be correlated with better collaborative task performance [15,28], common knowledge building [3], and learning gains [36,37] because it represents the language convergence on a shared understanding [14]. ...
Mathematical discussions have become a popular educational strategy to promote math literacy. While some studies have associated math literacy with linguistic factors such as verbal ability and phonological skills, no studies have examined the relationship between linguistic synchrony and math literacy. In this study, we modeled linguistic synchrony and students' math literacy from 20,776 online mathematical discussion threads between students and facilitators. We conducted Cross-Recurrence Quantification Analysis (CRQA) to calculate linguistic synchrony within each thread. The statistical testing result comparing CRQA indices between high and low math literacy groups shows that students with high math literacy have a significantly higher Recurrence Rate (RR), Number of Recurrence Lines (NRLINE), and the average Length of lines (L), but lower Determinism (DET) and normalized Entropy (rENTR). This result implies that students with high math literacy are more likely to share common words with facilitators, but they would paraphrase them. On the other hand, students with low math literacy tend to repeat the exact same phrases from the facilitators. The findings provide a better understanding of mathematical discussions and can potentially guide teachers in promoting effective mathematical discussions.
... les contraintes) fournissent un support probabiliste pour des interprétations compétitives en parallèle tout au long du temps et qu'elles auraient un poids différent dans la compréhension du langage. Les connaissances privées et les connaissances partagées influenceraient ainsi de façon distincte l'interprétation du langage en fonction de leurs poids dans la situation de communication (Brennan & Hanna, 2009 ;Brown-Schmidt & Hanna, 2011 ;Brown-Schmidt, 2012), raison pour laquelle un comportement collaboratif et un comportement égocentrique ont pu être rapportés en situation de conversation. Le fait que dans certains cas la prise en compte du destinataire émerge dans les étapes plus tardives ne signifie pas que le traitement précoce est nécessairement égocentrique. ...
The process of accommodating or adapting to the needs and knowledge of our interlocutor in a conversation is called audience design (Clark & Murphy, 1982; Clark, 1996; Galati & Brennan, 2010). We also speak of perspective taking since interlocutors have to take into account the perspective of their interlocutor when they formulate their statements (Brown-Schmidt & Heller, 2018). To understand when and how the interlocutors use audience design means to understand how speakers take into account information that they believe they share with their interlocutor, in other words their common ground. Audience design implies the distinction between what is part of the common ground (ie, the information shared with a particular interlocutor and mutually recognized as such; Clark, 1996), and what is part of the private knowledge (ie, privileged ground) specific to each speaker and unknown to the interlocutor. The common ground includes contextual knowledge (on the current physical context or related to previous conversational exchanges) and encyclopedic knowledge (such as knowledge shared by a specific community, e.g., the residents of Aix-en-Provence) shared by the speaker and the listener and recognized by the two interlocutors to be known by each of them (Clark & Marshall, 1981). Thus, the information shared or not shared by the interlocutors at one point in the conversation would determine the type of expressions or utterances used by the speaker for his/her interlocutor. While it is clear that speakers and listeners adapt their linguistic behavior to their interlocutor in order to avoid and resolve situations of incomprehension, to what extent, and in which situation, audience design is involved is still a question under debate. In other words, when, and how, do we take into account the perspective of our interlocutor when we choose particular linguistic forms (e.g., referential expressions, prosodic forms)? The aim of this article is to frame the studies we conducted at the LPL as part of this debate. Several audience design models have been proposed in psycholinguistics (e.g., collaborative model, egocentric model, interactive alignment model, dual processing model, probabilistic model, memory-based model) to account for the idea that speakers would not do inferences about the knowledge of their interlocutor continuously. These models diverge on the hypothesis of a more or less systematic use of the common ground in the implementation of audience design. While the collaborative model (Clark, 1996) considers conversation as a constant and joint contribution of the speaker and the listener to the common ground, suggesting a continuous engagement of the audience design in conversation, other models focus on the more or less speaker’s egocentric behavior and on the use of less-demanding automatic mechanisms. The main criticism against the collaborative model challenges the fact that perspective taking via the common ground would require a distinct and cognitively costly representation of knowledge, beliefs and intentions of the interlocutor. The processing costs of such a system would be too high to allow a smooth conversation. The egocentric (e.g., Keysar et al., 2000), interactive alignment (e.g., Pickering & Garrod 2004), probabilistic (e.g., Brown-Schmidt & Hanna, 2011), and memory-based (e.g., Horton & Gerring, 2005) models aim to take into consideration this cognitive cost. The dual processing model (e.g., Bard et al., 2000) attempts to reconcile collaborative and egocentric models by proposing two types of cognitive processes that prevail for linguistic choices: the production of referential expressions that requires making inferences about the knowledge of the interlocutor would involve slow and cognitively demanding processes, while the production of phonetic variations that refers to the speaker’s own recent experience would result from more automatic and rapid processes. At the same time as updating these models, several studies have identified speaker-internal constraints (e.g., memory, executive control, theory of mind) and situational constrains (e.g., visual context) which are likely to influence the involvement of design audience. Except for the dual processing model of Bard et al., models of audience design are mainly based on studies focusing either on the way speakers produce referential forms when referring to the same object at different times and/or with different addresses, or on the way listeners interpret referential forms online, as measured by the monitoring of eye movements (e.g., Keysar et al., 2000). Unlike these studies, our approach consists in studying the prosodic choices (i.e., the melodic and rhythmic choices) the speaker makes during conversation according to the presence of an interlocutor and the knowledge of this interlocutor.The aim of the current article is to describe: (1) the main psycholinguistic models contributing to understand when and how audience design takes place during conversation, and (2) to highlight the work we developed at the LPL in which we adopt a clinical approach to the study of audience design by evaluating the prosodic choices made by individuals with schizophrenia in relation to their difficulties in attributing mental states to others. More generally, our work with healthy participant shows that speakers 1) make different prosodic choices to indicate the relevant part of their message when speaking in the presence of a real interactional partner and when speaking without interlocutor 2) adapt their overall prosody relative to the visual context they share with the interlocutor to facilitate communication.
... However, although thanks to the use of algorithms and data structures, communication protocols between digital systems have been widely developed [36], communication between humans and robots is more complex due to the numerous elements of verbal and non-verbal signals that humans use [20] and their ability to convey emotions with language [25]. For the implementation of communication protocols using language, three elements must be considered: what is said, how it is said, and to whom it is said [37]. In social robots, what is said is programmed into the script, while how it is said is managed by various modules of transmission protocols [36]. ...
In recent years, in response to the effects of Covid-19, there has been an increase in the use of social robots in service organisations, as well as in the number of interactions between consumers and robots. However, it is not clear how consumers are valuing these experiences or what the main drivers that shape them are. Furthermore, it is an open research question whether these experiences undergone by consumers can be affected by their own personality. This study attempts to shed some light on these questions and, to do so, an experiment is proposed in which a sample of 378 participants evaluate a simulated front-office service experience delivered by a social robot. The authors investigate the underlying process that explains the experience and find that cognitive-functional factors, emphasising efficiency, have practically the same relevance as emotional factors, emphasising stimulation. In addition, this research identifies the personality traits of the participants and explores their moderating role in the evaluation of the experience. The results reveal that each personality trait, estimated between high and low poles, generates different responses in the evaluation of the experience.
... Not only will new objects with new properties emerge, but the way humans talk about objects changes over time. As has been shown repeatedly in experiments on human-human interaction, this is not only a long-term issue, but the exact meaning of language may often be negotiated and evolve during the course of a single interaction, and then develop into partner-specific language use (Brennan & Clark, 1996;Barr & Keysar, 2002;Brennan & Hanna, 2009;Ibarra & Tanenhaus, 2016;. This phenomenon has been referred to as conceptual pacts (Brennan & Clark, 1996), or more generally as alignment in communication (Pickering & Garrod, 2006). ...
This paper presents CoLLIE: a simple, yet effective model for continual learning of how language is grounded in vision. Given a pre-trained multimodal embedding model, where language and images are projected in the same semantic space (in this case CLIP by OpenAI), CoLLIE learns a transformation function that adjusts the language embeddings when needed to accommodate new language use. This is done by predicting the difference vector that needs to be applied, as well as a scaling factor for this vector, so that the adjustment is only applied when needed. Unlike traditional few-shot learning, the model does not just learn new classes and labels, but can also generalize to similar language use and leverage semantic compositionality. We verify the model’s performance on two different tasks of identifying the targets of referring expressions, where it has to learn new language use. The results show that the model can efficiently learn and generalize from only a few examples, with little interference with the model’s original zero-shot performance.
... Siamangs (Hylobates syndactylus) show stable patterns of note types that are consistent to a particular pair, in their duets and when repaired, the new pair forms a novel pair-specific duet (Geissmann 1999). In humans (Homo sapiens) partners converge or complement each other in their linguistic patterns and non-verbal behavior to form stable pair-specific communication norms (Brennan and Hanna 2009). ...
Social animals use complex communication to maintain social bonds. For pair-bonded animals in particular, acoustic communication is vital to reinforcing bonds and coordinating various behaviors within the bonded pair. Pair consistency in duets can be influenced by the time since pairing and may serve multiple functions; two of note are pair-bond reinforcement and advertisement of the relationship to conspecifics. Here, we assess the potential for pair consistency and plasticity in timing of vocalizations in 50 wild Northern gray gibbons (Hylobates funereus) from seven sites in Sabah, Malaysia. Specifically, we looked at the timing of the female gibbon’s great call and the subsequent male coda, the male duet contribution that follows the female great call (coda timing), and the duration of the male coda (coda duration). We found that for coda timing, pair-level variance was the most important source of variance, as opposed to intra-individual or inter-site variance. In contrast, for coda duration, individual-level variance was the most important source of variance. We also found that variability in the duration of female calls was not correlated with the timing or duration of male codas. Our results are consistent with previous work showing pair consistency in other paired, duetting species and contribute to the growing body of literature indicating that primate vocalizations—rather than being inflexible and innate—have a high degree of plasticity. Future work should aim to understand both how gibbons coordinate their duets and what impact this coordination may have on quality of the pair’s bond.
Significance statement
Duetting is seen across a diverse range of taxa, but in only a few nonhuman primates including indris, tarsiers, titi monkeys, and gibbons. Duetting in primates co-occurs with a suite of behavioral traits that include territoriality and pair bonding. Although the functions of primate duets remain a topic of debate, it is clear that duets provide information to neighboring conspecifics about the calling animal(s). Our results indicate that gibbon duets may contain information about pair identity. Investigating how duration of the pair bond influences either pair-level signatures or consistency of male timing will be an important next step in understanding the proximate influences that shape gibbon duet structure.
... Another view argues that information about an interlocutor's beliefs and perspective plays a more pivotal role in language processing, being one of multiple cues that help to constrain the interpretation of unfolding speech, including the processing of repeated labels (e.g. Brennan & Hanna, 2009;Brown-Schmidt & Hanna, 2011;Hanna et al., 2003). In response to Barr and Keysar (2002), Metzing and Brennan (2003) claimed that while a previously used term is indeed a strong cue that can facilitate access to a referent, regardless of the speaker, listeners may nevertheless track information about who produces certain labels, and use that information to guide processing. ...
Once interlocutors settle on a specific label in conversation, they tend to maintain the linguistic precedent and reuse the same label (i.e. they become lexically entrained). This helps to facilitate comprehension, with listeners identifying referents more quickly when repeated labels are used compared to new labels. In the current study, we looked at whether listeners are additionally sensitive to repeated infelicitous labels (Experiment 1), as when non-native speakers, for example, overgeneralise a term (e.g. identifying a chair as the chair with tires). In addition, we investigated the extent to which listeners’ expectations of incorrect labels are influenced by knowledge of community speaking patterns, testing whether listeners could disregard recently encountered lexical errors from a non-native speaker as possible labels when processing a native speaker, who should not be expected to produce such errors (Experiment 2). Our results provide no evidence that listeners were able to take into account speaker information.
... Interdisciplinary research has systematically demonstrated how our body shapes our perceptions, affording or constraining the experiences we entertain when relating to others (Uithol & Gallese, 2015), especially in situations of emotion recognition, joint attention, and action understanding (Reddy & Uithol, 2015). For example, adults coordinate in taking turns via nonverbal communication, evidencing the need for verbal synchronicity during dialogue (Brennan & Hanna, 2009;Shockley et al., 2003), and visually coordinate their attention through synchronized eye movements to understand each other's actions when completing joint actions (Schneider & Pea, 2014). This coordination does not necessarily require any sophisticated skills, even when cognitive systems are involved. ...
The
social distancing imposed by the pandemic transformed how people interact with others, and little is known about how it has impacted new ways of sociability and if culture influences this process. This is a qualitative study exploring changes in the configurations of social interactions and the resources for sociability that adults in Brazil, the USA, and Finland have developed during the initial stage of quarantine. A total of 95 participants (ages between 20 and 60) experiencing social isolation either living alone or with their partners (without children) completed online questionnaires about their interactive experiences. The questionnaire was composed of multiple choices, addressing the frequency, types, and length of social interactions before and during the pandemic, and open questions focusing on the participants’ experiences on online interactions during the pandemic. Frequencies were analyzed through a paired-sample t-test, and open-ended responses were thematically analyzed. Results revealed, first, that social isolation did not represent a significant change in the composition of the participants’ social network, but family bonds became the main connection during the period, and other sources of social interaction were kept due to the possibility of interaction through virtual means. Although the frequency of social interactions reduced, their significance increased. Second, virtual environments reframed social interactions, influencing individual’s bodily perceptions such as differences in attentional demands, communication processes, and awareness of their own image, and the interaction itself. Third, cultural values seemed to influence the way participants signified their interactive experiences. This study suggests that although virtual environments changed the ways interactions happen, virtual encounters were essential for maintaining participants’ social networks.
... Their use is more automated and nonconscious than the use of content words such as nouns and verbs (Segalowitz & Lane, 2004). Linguistically, function words reflect the relationship between content words and thereby signal shared knowledge among interaction partners (Brennan & Hanna, 2009;Cannava & Bodie, 2017). Each individual's specific pattern of function word use is called language style (Pennebaker & King, 1999;Pennebaker et al., 2003). ...
This article presents a dynamic conceptualization for the assessment of language style matching (LSM) over time. LSM is a team's mutual adaption of function words like pronouns, articles, or prepositions. LSM is a nonconsciously but frequently occurring communication behavior allowing researchers unobtrusive insights into teams' internal dynamics. Building on guidelines for the alignment of construct and measurement, a dynamic conceptualization and method for LSM are introduced. Simulated examples and interactions of N = 160 individuals in 26 teams indicate that dynamic LSM allows for a truer estimation of LSM than the hitherto used static method. Implications for future application are discussed.
... Second, if the audience tuning primarily serves conversational alignment with the audience (e.g., Brennan & Hanna, 2009), the memory bias might be due to greater cognitive effort invested in the audience-tuned information. We would, then, expect to find the audience-congruent memory bias in all conditions in which participants tuned their message to the audience. ...
General Audience Summary
Experts are often perceived as competent but cold. Their epistemic competence should qualify them as reliable sources in opinion formation, but their relational coldness should make them less desirable sources. To form judgments about an ambiguous or disputed topic, we can draw on the opinions of our communication partners and thus create a shared reality with them. Through shared-reality creation our judgments on a topic are aligned with the partner’s view. Importantly, people not only strategically adapt their communication to a partner’s opinion in this process, but actually change their own knowledge and opinion about the topic. This is shown by an evaluative bias in people’s subsequent memory for statements on an ambiguous topic that they had previously communicated about to an audience with a known opinion on the topic. Not only is the original message to the partner aligned with the audience’s opinion, but so too is the communicator’s own subsequent memory for the original information. Here, we examined whether people are more likely to create such a shared reality with an audience who is an expert on the communication topic or a non-expert. Previous research has shown that the creation of a shared reality is critically driven by two different motives: We want to feel socially close to others (relational motive), and we want to know what is true (epistemic motive). In two studies, we found that shared reality creation, as indicated by the memory bias toward the audience’s opinion, was created when experts rather than non-experts formed the audience. This effect was not simply due to expert’s higher social status because a highly educated, high-status audience who lacked specific expertise on the topic was treated like a lay audience. These findings are critical to science communication and the influence experts can have on public opinion.
... Linguistic adaptation has been observed on multiple levels of linguistic representations, such as phonetic and phonological processing (Clayards, Tanenhaus, Aslin, & Jacobs, 2008;Kleinschmidt & Jaeger, 2015;Kraljic, Samuel, & Brennan, 2008;Maye, Aslin, & Tanenhaus, 2008;Norris, McQueen, & Cutler, 2016), sentence level syntactic processing (Fine et al., 2013;Jaeger & Snider, 2013;Kaschak & Borreggine, 2008;Kaschak & Glenberg, 2004), and discourse context (Ryskin et al., 2020;Schuster & Degen, 2020;Van Berkum et al., 2005). Flexibility in adapting to speaker-specific information has been also observed in the perception of differences in pronunciation such as foreign accents and dialects (Bradlow & Bent, 2008;Maye et al., 2008), lexical choice (Brennan & Hanna, 2009), and the role of prosody in making pragmatic inferences (Kurumada, et al., 2014). ...
A growing body of research suggests that language users integrate diverse sources of information in processing and adapt to the variability of language at multiple levels. In two visual-world paradigm studies, we explored whether listeners use prosody to predict a resolution to structures with a PP that is structurally ambiguous between a modifier and an instrument interpretation. The first study revealed that listeners predict a referent that is most compatible with the location of a prosodic boundary, casting anticipatory looks to the appropriate object even before the onset of a disambiguating word. The second study indicated that listeners failed to anticipate instrument resolutions when the prosody of non-experimental filler items was unconventional, even though experimental items remained identical to the first study. The results suggest that listeners adjust their predictive processing to the utility of prosodic information according to whether a speaker reliably conforms to the conventional use of prosody.
... The classical versions of the residual activation and implicit learning accounts of structural priming do not build in speaker-specific representations; as such, priming from a conversation with one speaker would be expected to carry over to any subsequent language use. However, there is now a good deal of evidence that individuals track usage properties of specific speakers (Brennan & Hanna, 2009;Kleinschmidt & Jaeger, 2015;Pogue et al., 2016;Yildirim et al., 2016). This raises the question of whether speakers generalise the structural usage patterns of one interlocutor to a new interlocutor. ...
This study extends the logic of prior studies showing phonetic convergence between interlocutors to the structural domain. We ask whether listeners’ adaptation of the syntactic forms they produce depends on their perceptions about their interlocutor's social proximity and linguistic competence, using structural priming as a measure of convergence. Two experiments compared structural priming in dialogues between native British English speakers and (i) other native British English speakers, (ii) native speakers of North American English, and (iii) non-native speakers of English, to assess to what extent interlocutor characteristics influence structural convergence in dialogue. Our findings suggest that rates of structural convergence depend both on a speaker's pre-existing structural biases for particular verbs, and their perception of (linguistic or social) similarity to their interlocutor. This suggests that low-level mechanisms underlying structural convergence may be mediated by beliefs about how interlocutors are socially situated with respect to each other.
... However, fewer studies focused on the beneficial effect of adapting one's behavior in social interaction. The literature on the subject comes mostly from the communication (Brennan and Hanna, 2009) and the medical interaction fields (Kiesler and Auerbach, 2006) and mainly focuses on the adaptation of verbal behaviors. In comparison, the adaptation of non-verbal behaviors (NVBs) has been scarcely investigated. ...
Human diversity cannot be denied. In our everyday social interactions, we constantly experience the fact that each individual is a unique combination of characteristics with specific cultural norms, roles, personality, and mood. Efficient social interaction thus requires an adaptation of communication behaviors to each specific interlocutor that one encounters. This is especially true for non-verbal communication that is more unconscious and automatic than verbal communication. Consequently, non-verbal communication needs to be understood as a dynamic and adaptive process in the theoretical modeling and study of social interactions. This perspective paper presents relevance, challenges, and future directions for the study of non-verbal adaptation in social interactions. It proposes that non-verbal adaptability is more pertinently studied as adaptation to interlocutor's inner characteristics (i.e., expectations or preferences) than to interlocutor's behaviors per se, because behaviors are communication messages that individuals interpret in order to understand their interlocutors. The affiliation and control dimensions of the Interpersonal Circumplex Model are proposed as a framework to measure both the interlocutors' inner characteristics (self-reported) and the individuals' non-verbal responses (external coders). These measures can then be compared across different interactions to assess an actual change in behavior tailored to different interlocutors. These recommendations are proposed in the hope of generating more research on the topic of non-verbal adaptability. Indeed, after having gathered the evidence on average effects of non-verbal behaviors, the field can go further than a “one size fits all” approach, by investigating the predictors, moderators, and outcomes of non-verbal adaptation to the interlocutors' inner characteristics.
... While it is not easily classified into either the completepooling or no-pooling classes, it is also not straightforward for the priming mechanisms proposed by interactive alignment accounts to explain these patterns of partner-specificity without being augmented with additional social information. If a particular semantic representation has been activated due to precedent in the preceding dialogue, then the identity of the speaker should not in principle alter its continued influence (Brennan & Hanna, 2009). More sophisticated hierarchical memory retrieval accounts that represent different partners as different contexts (e.g Polyn, Norman, & Kahana, 2009) may be consistent with partner-specificity, but evoking such an account presupposes that social information like partner identity is already a salient and relevant feature of the communicative environment and thus no longer relies purely on "egocentric" priming and activation mechanisms. ...
Languages are powerful solutions to coordination problems: they provide stable, shared expectations about how the words we say correspond to the beliefs and intentions in our heads. Yet language use in a variable and non-stationary social environment requires linguistic representations to be flexible: old words acquire new ad hoc or partner-specific meanings on the fly. In this paper, we introduce a hierarchical Bayesian theory of convention formation that aims to reconcile the long-standing tension between these two basic observations. More specifically, we argue that the central computational problem of communication is not simply transmission, as in classical formulations, but learning and adaptation over multiple timescales. Under our account, rapid learning within dyadic interactions allows for coordination on partner-specific common ground, while social conventions are stable priors that have been abstracted away from interactions with multiple partners. We present new empirical data alongside simulations showing how our model provides a cognitive foundation for explaining several phenomena that have posed a challenge for previous accounts: (1) the convergence to more efficient referring expressions across repeated interaction with the same partner, (2) the gradual transfer of partner-specific common ground to novel partners, and (3) the influence of communicative context on which conventions eventually form.
... From the perspective of the communication during interactions, it is also relevant to mention that studies have shown that adults coordinate in taking turns via nonverbal communication, evidencing the need for verbal synchronicity during collaborative dialog (Brennan & Hanna, 2009), as well as how participants in dyads sway their bodies in synchronized ways, especially when conversing with each other (Shockley et al., 2003), and visually coordinate their attention through synchronized eye movements to understand each other's actions when completing collaborative tasks in pairs (Schneider & Pea, 2014). These studies showed how the body is also used to establish the synchrony necessary for collaborative interactions, demonstrating that is not at all clear yet what are the boundaries between speech and movement, and to what direction this relation is established-i.e., in speech influencing movement entrainments or vice versa. ...
The bodily experiences and implications of understanding the functioning of the human brain–body mechanism has been a center of attention in the field of cognitive neurosciences for over two decades. Research in this field has enlarged the theories of learning and development, and contributed to changes in educational practices involving language processing, mathematics, and spatial thinking; however, these changes have not yet been applied to the analysis of transversal competencies such as collaborative learning. The aim of this paper is to bridge the theoretical and applied advances in the field of embodied cognition, specifically collaborative learning. The definitions, theoretical frameworks, and current methodological approaches in the field of collaborative learning are reviewed, with a particular focus on those studies that have investigated interactive dynamics in collaborative situations. The need to take the field further by exploring the theoretical perspective of embodied cognition as a possibility that can open the field is also presented. The relevance of investigating learning in groups by analyzing bodily engagements and intersubjectivity is demonstrated and methodological considerations are raised.
... L'émergence de tels phénomènes de coopération serait la conséquence évolutive d'un besoin d'adaptation aux perpétuels mouvements d'un environnement toujours plus riche et complexe (voir Tomasello, 2009 La cognition sociale fait référence à l'ensemble des processus cognitifs qui supportent notre perception et compréhension d'autrui, ainsi que les interactions que nous entretenons avec ce dernier (Gallotti & Frith, 2013). En tant que concept général, la cognition sociale renvoie à des compétences aussi variées que l'empathie (Melloni, Lopez, & Ibanez, 2013), l'imitation (Chartrand & Lakin, 2013), la prise de perspective d'autrui (Quesque, Chabanat, & Rossetti, 2018), la communication verbale (Brennan & Hanna, 2009) et sensori-motrice (Pezzulo, Donnarumma, & Dindo, 2013), la coordination interpersonnelle (Semin & Cacioppo, 2008) ou encore la théorie de l'esprit (Goldman & De Vignemont, 2009). En situation d'interaction, ces différentes compétences s'organisent conjointement afin de nous permettre de comprendre et anticiper avec précision les actions qu'autrui s'apprête à réaliser. ...
Il est communément accepté que le simple fait d’agir à proximité d’autrui nous amène naturellement à considérer son action, et ses potentielles conséquences sur l’environnement. L’une des principales conséquences liée à cette tendance à (co-)représenter les actions d’autrui se caractérise par le besoin de discriminer entre des représentations référant à soi et à autrui. Malgré des avancées majeures, aujourd’hui encore, des oppositions théoriques et mécanistiques demeurent. Ces dernières se cristallisent autour de l’interprétation de l’incidence de certains facteurs censés affecter la genèse et la force de ce problème de discrimination. En cela, la nature de ce phénomène, sa résolution, et ce à quoi réfère vraiment la mesure typiquement utilisée pour l’illustrer sont discutés tout du long de ce document. Ce travail de thèse soutient une perspective incarnée de la cognition sociale en démontrant le rôle déterminant des expériences sensori-motrices, et du codage redondant des informations qui en résultent, dans le processus de résolution du problème de discrimination. Nous faisons ainsi l’hypothèse que la mesure utilisée est avant tout un indicateur de la facilité avec laquelle nous résolvons le problème de discrimination entre soi et autrui.
... There is a debate in the psycholinguistics community about whether convergence of communicative behaviours derives from partner adaptation or from recency adaptation (Brennan 1996;Brennan and Hanna 2009;Brennan et al. 2013;Stenchikova and Stent 2007). Partner adaptation refers to an adaptation of behaviour based on a model of the partner. ...
Alignment of communicative behaviour is an important feature of Human–Human interaction that directly affects the collaboration and the social connection of conversational partners. With the aim of improving the communicative abilities of a virtual agent, and in particular its strategies related to (lexical) verbal alignment, this article focuses on the alignment of linguistic productions of dialogue participants in task-oriented dialogues. We propose a new framework to quantify both the lexical alignment and the self-repetition behaviours of dialogue participants from dyadic dialogue transcripts. The framework involves easily computable measures based on repetition of lexical patterns automatically extracted via a sequential pattern mining approach. These measures allow the characterisation of the nature of these processes by addressing various informative aspects such as their variety, complexity, and strength. This framework is implemented in the freely available and open-source software dialign. Using these measures, we present a contrastive study between Human–Human and Human–Agent dialogues on various corpora that reveals major differences in the lexical alignment and self-repetition behaviours. Lastly, we address the challenge of integrating lexical alignment capabilities in artificial agents. To this end, we describe guidelines and we discuss the integration of the proposed framework in a real-time dialogue system.
... Anti-mentalistic models of conversation consequently relegate mindreading to a minor, supporting role, for example, as a device for repairing certain kinds of misunderstandings (Apperly, 2018;Pickering & Garrod, 2004;Shintel & Keysar, 2009). These proposals are challenged, however, by a wide array of evidence indicating that speakers are sensitive to information about speakers' mental states from the very earliest stages of language production and comprehension (Brennan & Hanna, 2009;Hanna et al., 2003;Heller et al., 2012;Rubio-Fernández et al., 2019). It is not obvious how to reconcile this latter research with the evidence suggesting that mindreading is slow and cognitively demanding. ...
How is human social intelligence engaged in the course of ordinary conversation? Standard models of conversation hold that language production and comprehension are guided by constant, rapid inferences about what other agents have in mind. However, the idea that mindreading is a pervasive feature of conversation is challenged by a large body of evidence suggesting that mental state attribution is slow and taxing, at least when it deals with propositional attitudes such as beliefs. Belief attributions involve contents that are decoupled from our own primary representation of reality; handling these contents has come to be seen as the signature of full-blown human mindreading. However, mindreading in cooperative communication does not necessarily demand decoupling. We argue for a theoretical and empirical turn towards "factive" forms of mentalizing here. In factive mentalizing, we monitor what others do or do not know, without generating decoupled representations. We propose a model of the representational, cognitive, and interactive components of factive mentalizing, a model that aims to explain efficient real-time monitoring of epistemic states in conversation. After laying out this account, we articulate a more limited set of conversational functions for nonfactive forms of mentalizing, including contexts of meta-linguistic repair, deception, and argumentation. We conclude with suggestions for further research into the roles played by factive versus nonfactive forms of mentalizing in conversation.
... In the context of this experimental design, the non-monotonic pattern of the influence of speaker is inconsistent with a onestage process that would reflect the functioning of a constraintbased mechanism (Brown-Schmidt, 2009;Brennan and Hanna, 2009;Brown-Schmidt and Hanna, 2011). Basically, the two sources of information that could be used to disambiguate the referential expression "the bat"-the presence of a precedent and speaker identity-pointed to the same alternative, in our example, the mammal bat. ...
To reduce ambiguity across a conversation, interlocutors reach temporary conventions or referential precedents on how to refer to an entity. Despite their central role in communication, the cognitive underpinnings of the interpretation of precedents remain unclear, specifically the role and mechanisms by which information related to the speaker is integrated. We contrast predictions of one-stage, original two-stage, and extended two-stage models for the processing of speaker information and provide evidence favoring the latter: we show that both stages are sensitive to speaker-specific information. Using an experimental paradigm based on visual-world eye tracking in the context of a referential communication task, we look at the moment-by-moment interpretation of precedents and focus on the temporal profile of the influence of the speaker and linguistic information when facing ambiguity. We find two clearly identifiable moments where speaker-specific information has its effects on reference resolution. We conclude that these two stages reflect two distinct cognitive mechanisms, with different timings, and rely on different representational formats for encoding and accessing information about the speaker: a cue-driven memory retrieval process that mediates language processing and an inferential mechanism based on perspective-taking abilities.
... For instance, a piece of information i produced during an interaction between Speaker A and Speaker B is deemed part of their common ground by both speakers. Dialogue partners then resort to their common ground to produce utterances that are adapted to their current partner, ensuring mutual comprehension (Achim et al., 2017;Brennan & Clark, 1996;Brennan & Hanna, 2009;Fussell & Krauss, 1992). For instance, because i belongs to A and B's common ground, either speaker would suppose that their partner is capable of easily understanding an utterance about i. ...
Conversational memory is subject to a number of biases. For instances, references which were reused during dialogue are remembered better than non-reused references. Two experiments examined whether speakers are aware that they are subject to such biases and whether they use information about reference origin (i.e., information about who said what) to determine which references are remembered better by their partner. Pairs of participants performed a map task followed by a questionnaire that assessed each participant's content memory as well as each participant's estimation of his or her partner's memory. In Experiment 1, the participants were unaware that they would perform a memory test after the map task, whereas the participants were mutually aware of the upcoming memory test in Experiment 2. The results revealed that participants did know that their partner was subject to memory biases and that their estimation of these biases was mainly accurate. The results prevented us from drawing any conclusions about the role of origin memory in this process. These results have important implications for subsequent dialogic partner-adaptation.
... Furthermore, this characterization of icg is meant to follow Horton and Gerrig's main motivation, namely to integrate Common Ground mechanisms. In this sense, icg is meant to be determined by mechanisms that have been proven to be relevant to conversational dynamics, such as attention allocation (Wardlow and Ferreira 2008;Brennan and Hanna 2009), and allows to take into account factors like location change (Meagher and Fowler 2014). ...
Classical notions of Common Ground have been criticized for being cognitively demanding given their appeal to complex meta-representations. The authors here propose a distinction between Immediate Common Ground, containing information specific to the communicative situation, and General Common Ground, containing information that is not situation-specific. This distinction builds on previous work by Horton and Gerrig [2016], extending the idea that common cognitive processes are part of the establishment and use of common ground. This is in line with the idea that multiple cognitive resources are involved in dialogue and avoids appealing to special-purpose representations for Common Ground purposes.
This study investigates whether interpersonal coordination of language style in written text message communication relates to past-year depressive symptoms and lifetime major depressive disorder (MDD) in young adults. Consistent with application of Joiner's integrative interpersonal framework to interpersonal coordination, we hypothesized that students with more experiences of depression, and their conversation partners, would engage in less interpersonal coordination in text messages (indexed by reciprocal language style matching of function words; rLSM). College students at a large southeastern university ( N = 263) contributed two weeks of text messages in 2014−2015, alongside a mental health survey. Texts were filtered to dyads that used formal English (207,942 talk turns), accommodating limitations of LSM measurement. Structural equation models showed that students with more past-year depressive symptoms and lifetime MDD coordinated more (opposite the hypothesized direction of effect). Implications for interpersonal processes in depression and measurement of rLSM in text messages are discussed.
Inspired by early proposals in philosophy, dominant accounts of language posit a central role for mutual knowledge, either encoded directly in common ground, or approximated through other cognitive mechanisms. Using existing empirical evidence from language and memory, we challenge this tradition, arguing that mutual knowledge captures only a subset of the mental states needed to support communication. In a novel theoretical proposal, we argue for a cognitive architecture that includes separate, distinct representations of the self and other, and a cognitive process that compares these representations continuously during conversation, outputting both similarities and differences in perspective. Our theory accounts for existing data, interfaces with findings from other cognitive domains, and makes novel predictions about the role of perspective in language use. We term this new account the Multiple Perspectives Theory of mental states in communication.
To provide an engaging and natural interaction with an argumentative dialogue system, we introduce a model to adapt the system utterances to the user’s communication style in an ongoing discussion. Therefore, we propose an “Elaborateness Score” (ES) considering the length of the user utterances and requested content and adapts its utterance length as well as the amount of provided meta-information accordingly. To evaluate our approach we conducted a laboratory user study with a total of 30 participants who had a conversation with our ADS. In a pre- and post-survey, the participants had to rate statements to subjectively assess each dialogue and indicate which level of elaborateness they preferred as well as their general impression of the system. The results show that the system’s elaborateness style has a significant influence on user’s perception of the dialogue and imply that the preference of the system’s elaborateness is individual for every user.KeywordsElaborateness ScoreCommunication StylesHuman-Computer InteractionSpoken Dialogue SystemsCooperative Argumentative Dialogue Systems
The present study provides converging evidence across three next-mention biases that likelihood of coreference influences the choice of referring expression: Implicit Causality (IC), the goal bias of Transfer-of-Possession verbs, and Implicit Consequentiality. A pilot study and four experiments investigated coreference production in German using a forced-reference paradigm (Fukumura and van Gompel 2010). The pilot study used object- and subject-biased IC verbs, showing a statistically marginal influence of next-mention bias on referential expressions, albeit mediated by grammatical function and feature overlap between antecedents. Experiment 1 focused on these features for object reference with Transfer-of-Possession verbs (Rosa and Arnold 2017), showing effects of coreference bias. In a within- participants comparison, Experiment 2 showed comparable effects for two classes of IC verbs, stimulus-experiencer and experiencer-stimulus predicates. Experiment 3 replicated and extended the IC form effects to another verb class, agent-evocator verbs. Finally, Experiment 4 revealed effects on anaphoric form also for Implicit Consequentiality, while simultaneously replicating the effect observed for IC.
In this article, we aim to evaluate the role of robots’ personality-driven behavioural patterns on users’ intention to use in an entertainment scenario. Toward such a goal, we designed two personalities: one introverted with an empathic and self-comparative interaction style, and the other extroverted with a provocative and other-comparative interaction style. To evaluate the proposed technology acceptance model, we conducted an experiment ( N = 209) at a public venue where users were requested to play a game with the support of the TIAGo robot. Our findings show that the robot personality affects the acceptance model and three relevant drivers: perceived enjoyment, perceived usefulness, and social influence. The extroverted robot was perceived as more useful than the introverted, and participants who interacted with it were faster at solving the game. On the other hand, the introverted robot was perceived as more enjoyable but less useful than the extroverted, and participants who interacted with it made fewer mistakes. Taken together, these findings support the importance of designing proper robot personalities in influencing users’ acceptance, featuring that a given style can elicit a different driver of acceptance.
According to the communicative efficiency hypothesis, speakers should produce more linguistic material when comprehension difficulty increases. Here, we investigate a potential source of comprehension difficulty – listeners’ language proficiency – on speakers’ productions, using referential choice as a case study. Referential choice is influenced by communicative efficiency: pronouns are used less than full noun phrases (NPs) for less predictable referents (Tily & Piantadosi, 2009). However, the extent to which it is influenced by the listener is debated. Here, we compare participants’ descriptions of the same picture book to children, adult L2 learners and adult native speakers. We find that speakers use more full NPs when their interlocutors are learners – child and adult learners alike, illustrating an effect of listeners’ proficiency (regardless of age) on production choices. Importantly, the increased use of full NPs relative to pronouns is found controlling for discourse-related differences (e.g., previous mention), suggesting a direct relation between listeners’ perceived language proficiency and referential choice.
Variety contact – the communication of speakers of different language varieties – is a phenomenon that has increased recently not only in the German-speaking world, but in other languages as well. New forms of digital communication and new transnational possibilities of migration within Europe have, among other factors, contributed to the fact that speakers experience variety contact in their everyday life. Variety usage displays a great deal of information about speakers and it affects the outcome of ongoing interactions with others. Varieties are not explicitly learned, e. g. through language courses, thus variety-contact interactions are the only possibilities for learning and adapting to other varieties. In comparison to classical language learning settings, these situations provide only little explicit feedback. In particular, the learning of syntax is often associated with critical periods. The learning process of adults regarding new syntactic variants in variety-contact interactions is regarded as difficult. A variety-contact experiment was developed and set-up with 30 participant pairs as a confederate referential communication task to investigate these issues. Prior to that, a cognitive variety concept was developed in the enactive paradigm that is compatible with the learning processes of autonomous, autopoietic, learning speakers. A typical Bavarian variant, doubly-filled COMP relative clauses, was chosen as a stimulus, as the results of a corpus analysis of the oral Zwirner Data and an online questionnaire show. To ensure that participants noted the syntactic variant in the contact interactions, an online perception experiment was conducted with 435 participants from Austria, Germany and Switzerland. It revealed a high grade of saliency for the phenomenon. In the contact experiment standard speakers from northern Germany that migrated recently to Austria were brought in contact with Bavarian speakers from Austria. They played a computer-mediated language game that evoked the usage of relative clauses with changing roles. The confederate Bavarian speakers were trained to use doubly-filled COMP relative clauses such that the Northern Germans’ behaviour could be observed regarding accommodation processes. None of them used unambiguous instances of the phenomenon during the experiment, but half of them did in a subsequent imitation task. This shows that these speakers learned the phenomenon at least perceptively and could use it as a socio-indexical variant to signal the Bavarian dialect in the production task. It was also shown that bad scores for the acceptability of that phenomenon increased because of the contact interaction. Especially those participants who adapted more strongly to the syntactic variant a) had higher personality scores with respect to “openness” and “conscientiousness” and b) had been in Austria for about 6 months longer than the other group. This indicates that repetitive variety contact may lead to variety learning even under difficult learning conditions and even for the difficult case of syntax learning of adults. - URL: https://utheses.univie.ac.at/detail/49288#
Languages are powerful solutions to coordination problems: They provide stable, shared expectations about how the words we say correspond to the beliefs and intentions in our heads. Yet, language use in a variable and nonstationary social environment requires linguistic representations to be flexible: Old words acquire new ad hoc or partner-specific meanings on the fly. In this article, we introduce continual hierarchical adaptation through inference (CHAI), a hierarchical Bayesian theory of coordination and convention formation that aims to reconcile the long-standing tension between these two basic observations. We argue that the central computational problem of communication is not simply transmission, as in classical formulations, but continual learning and adaptation over multiple timescales. Partner-specific common ground quickly emerges from social inferences within dyadic interactions, while community-wide social conventions are stable priors that have been abstracted away from interactions with multiple partners. We present new empirical data alongside simulations showing how our model provides a computational foundation for several phenomena that have posed a challenge for previous accounts: (a) the convergence to more efficient referring expressions across repeated interaction with the same partner, (b) the gradual transfer of partner-specific common ground to strangers, and (c) the influence of communicative context on which conventions eventually form. (PsycInfo Database Record (c) 2022 APA, all rights reserved).
Scalar implicatures involve inferring the use of a less informative term (e.g. some) to mean the negation of a more informative term (e.g. not all). A growing body of recent research on the derivation of scalar implicatures by adult second language (L2) learners shows that while they are successful in acquiring the knowledge of scalar implicatures, a property at the semantics–pragmatics interface, it remains controversial as to which mechanism, default or non-default, could account for L2 learners’ derivation of scalar implicatures. The present study used an online self-paced reading task to address this issue by examining the role of the speaker’s knowledge state in the interpretation of the existential quantifier some by Chinese-speaking learners of English in incremental sentence processing. Results showed that both L2 and native participants demonstrated comparable online sensitivity to the speaker’s knowledge state. Critically, when the scalar implicature was computed in situations where the speaker was more likely to know whether the statement with the stronger alternative was true, it gave rise to measurable reading latency, indicative of increased processing costs. We conclude by arguing that our findings are compatible with the context-driven models within the Gricean tradition.
An ongoing debate in the interpretation of referring expressions concerns the degree to which listeners make use of perspective information during referential processing. We aim to contribute to this debate by considering perspective shifting in narrative discourse. In a web-based mouse-tracking experiment in Dutch, we investigated whether listeners automatically shift to a narrative character’s perspective when resolving ambiguous referring expressions, and whether different linguistic perspective-shifting devices affect how and when listeners switch to another perspective. We compared perspective-neutral, direct, and free indirect discourse, manipulating which objects are visible to the character. Our results do not show a clear effect of the perspective shifting devices on participants’ eventual choice of referent, but our online mouse-tracking data reveal processing differences that suggest that listeners are indeed sensitive to the conventional markers of perspective shift associated with direct and (to a lesser degree) free indirect discourse.
Common ground can be mutually established between conversational partners in several ways. We examined whether the modality (visual or linguistic) with which speakers share information with their conversational partners results in memory traces that affect subsequent references addressed to a particular partner. In 32 triads, directors arranged a set of tangram cards with one matcher and then with another, but in different modalities, sharing some cards only linguistically (by describing cards the matcher couldn't see), some only visually (by silently showing them), some both linguistically and visually, and others not at all. Then directors arranged the cards again in separate rounds with each matcher. The modality with which they previously established common ground about a particular card with a particular matcher (e.g., linguistically with one partner and visually with the other) affected subsequent referring: References to cards previously shared only visually included more idea units, words, and reconceptualizations than those shared only linguistically, which in turn included more idea units, words, and reconceptualizations than those shared both linguistically and visually. Moreover, speakers were able to tailor references to the same card appropriately to the distinct modality shared with each addressee. Such gradient, partner-specific adaptation during re-referring suggests that memory encodes rich-enough representations of multimodal shared experiences to effectively cue relevant constraints about the perceptual conditions under which speakers and addressees establish common ground.
Originating in the 1960’s with the work of William Labov, the field of sociolinguistics has given way to a rich literature that continues to uncover the many ways in which social factors influence how we produce, perceive, and process speech. Sociolinguistic research has burgeoned alongside increasing globalization and migration, which has, in the case of the U.S. at least, resulted in increased levels of bilingualism and more frequent interactions with non-native English speakers. My dissertation, which consists of three distinct chapters, combines insights from the sociolinguistic literature with methodologies from cognitive science in order to better understand the ways in which perceptions of identity and social attitudes towards nonstandard language varieties influence our everyday spoken interactions. More specifically, I investigate how several social factors (i.e. language background, dialect stigmatization, and speaker accent) may influence speech production, perception, and processing. The data presented come from over sixty fieldwork interviews, a series of corpus analyses, two online surveys, and one neurolinguistic experiment. In the first paper, I identify how social factors have appeared to influence auxiliary verb choice among some Ecuadorian Spanish speakers. While the markedly frequent use of auxiliary ir, Sp. ‘to go’ in Ecuadorian Spanish has historically been traced to contact effects from Quichua, analysis of a present-day Ecuadorian Spanish corpus reveals that Quichua-Spanish bilinguals do not use the construction significantly more than Spanish monolinguals. Given auxiliary ir may be marked as a slightly nonstandard alternative for the auxiliary estar and that Quichua-Spanish bilinguals have long been denied linguistic prestige in the sociolinguistic stratification of Ecuadorian Spanish, I propose the possibility that language background and dialect stigmatization may explain the current distribution of auxiliary ir production among Ecuadorian Spanish speakers. In the second chapter, I investigate the relationship between speaker accents and American perceptions of nativeness. Specifically, I examined how young adult Midwesterners today perceive two main kinds of Spanish-influenced English varieties: L1 Latino English (as spoken in Chicago, U.S.) and L2 Spanish-accented English (as spoken in Santiago, Chile). Since Latinos have recently become the dominant ethnic minoritized group in the U.S., the varieties of English that they speak are under increasing scrutiny, and cases of linguistic discrimination are on the rise. Results from an accent evaluation survey reveal that respondents distinguished the L1 Latino English from the L2 Spanish-influenced English speaker, but still rated him as slightly more foreign-sounding than L1 speakers with more established U.S. dialects (e.g. New York). In other words, native U.S. speakers perceived as “sounding Hispanic” were perceived as sounding “almost American,” which suggests that what Midwesterners count as sounding American may be in the process of expanding to include U.S.-born Latinos. In the third chapter, I focus on the effect that speaker accent has on online word processing in the brain. Specifically, does Spanish-accented English speech increase activation of the Spanish lexicon in the mind of Spanish-English bilingual listeners? Though more data is needed for a clear answer, preliminary data from an EEG experiment suggests that speaker accent may possibly modulate bilingual lexical activation. This is investigated via analysis of N400 responses from bilingual listeners when false cognates from Spanish were produced by a Spanish-accented English speaker relative to a Chinese-accented English speaker.
When two people interact, reference presentation is shaped with the intention of supporting addressee understanding, allowing for ease of acceptance, thus minimizing overall collaborative effort. To date, analysis of such audience design has focused largely on adult–adult or adult–child interaction but seldom on adult–teenager interaction, including teacher–student interaction. An experiment was conducted in a British school in which teachers and students interacted to establish a reference for abstract tangram figures. Teachers were able to account for the students’ increased ability to behave in a more adult-like collaborative way with dialogue features similar to those in adult–adult contexts. Set apart was dialogue with young students, where teachers continued to guide the interaction by producing lengthier descriptions and by encouraging participation. Dialogue with young students differs from that with other teachers in terms of the amount of effort put into the interaction and in how this effort is distributed and shared among dialogue partners.
The study of human-human communication and the development of computational models for human-agent communication have diverged significantly throughout the last decade. Yet, despite frequently made claims of “super-human performance” in, e.g., speech recognition or image processing, so far, no system is able to lead a half-decent coherent conversation with a human. In this paper, we argue that we must start to re-consider the hallmarks of cooperative communication and the core capabilities that we have developed for it, and which conversational agents need to be equipped with: incremental joint co-construction and mentalizing. We base our argument on a vast body of work on human-human communication and its psychological processes that we reason to be relevant and necessary to take into account when modeling human-agent communication. We contrast those with current conceptualizations of human-agent interaction and formulate suggestions for the development of future systems.
When people in conversation refer repeatedly to the same object, they come to use the same terms. This phenomenon, called lexical entrainment, has several possible explanations. Ahistorical accounts appeal only to the informativeness and availability of terms and to the current salience of the object's features. Historical accounts appeal in addition to the recency and frequency of past references and to partner-specific conceptualizations of the object that people achieve interactively. Evidence from 3 experiments favors a historical account and suggests that when speakers refer to an object, they are proposing a conceptualization of it, a proposal their addressees may or may not agree to. Once they do establish a shared conceptualization, a conceptual pact, they appeal to it in later references even when they could use simpler references. Over time, speakers simplify conceptual pacts and, when necessary, abandon them for new conceptualizations.
Two pairs of studies examined effects of perspective taking in communication, using a 2-stage methodology that first obtained people's estimates of the recognizability to others of specific stimuli (public figures and everyday objects) and then examined the effects of these estimates on message formulation in a referential communication task. Ss were good at estimating stimulus identifiability but were biased in the direction of their own knowledge. The amount of information in a referring expression varied inversely with the perceived likelihood that addressees could identify the target stimulus. However, effects were less strong than anticipated. Although communicators do take others' knowledge into account, the extent to which they do so involves a trade-off with other sorts of information in the communicative situation.
In discourse, speakers tend to choose lexically short words (e.g., pronouns) when the words’ referents are highly accessible to listeners. However, in narrations of a film, a change in episode between references to a character, even one who should otherwise be accessible to a listener, tends to block use of short expressions. In one investigation of spontaneous film narrations and in two follow-up experiments, we found that conditions fostering shortening and lengthening at the lexical level also fostered durational reduction and blocking of reduction of repeated names and of content words more generally. The experiments confirm that episode boundaries tend to block durational shortening, but only when boundaries are marked by “metanarrative statements” (references, e.g., to a scene as such) not by narrative-level discontinuities.
Outlines features of interactive spoken discourse that distinguish it from other forms of discourse and the research methods that have evolved to examine it. Then this chapter reviews some of the basic facts about production and comprehension in conversation, which shows various ways that speakers and addressees take each other into account, or at least seem to. The authors present arguments from the discourse literature about about sorts of mental processes and representations that are needed to explain these findings. Based on these arguments, they lay out a set of theoretical distinctions that need to be kept in mind in the empirical study of conversation. Finally, with these distinctions in mind, this chapter reviews the burgeoning and controversial empirical literature that focuses on when and whether people adapt their discourse to each other. (PsycINFO Database Record (c) 2012 APA, all rights reserved)
Examined how native and non-native speakers adjust their referring expressions to each other in conversation. 20 Asian language speakers learning English were tested before and after conversations with native English speakers in which they repeatedly matched picture of common objects (Exp 1). Lexical entrainment was just as common in native/non-native pairs as in native/native pairs. People alternated director/matcher roles in the matching task; natives uttered more words than non-natives in the same roles. In Exp 2, 31 natives rated the pre- and post-test expressions for naturalness; non-natives' post-test expressions were more natural than their pre-test expressions. In Exp 3, 20 natives rated expressions from the transcribed conversations. Native expressions took longer to rate and were judged less natural-sounding when they were addressed to non-natives than to other natives. These results are consistent with H. H. Clark and D. Wilkes-Gibbs's (see record
1987-07185-001) principle of Least Collaborative Effort. (PsycINFO Database Record (c) 2012 APA, all rights reserved)
Natural languages contain probabilistic constraints that influence the resolution of ambiguities. Current models of sentence processing agree that probabilistic constraints affect syntactic ambiguity resolution, but there has been little investigation of the constraints themselves-what they are, how they differ in their effects on processing, and how they interact with one another. Three different types of probabilistic constraints were investigated: “pre-ambiguity” plausibility information, information about verb argument structure frequencies, and “post-ambiguity” constraints that arrive after the introduction of the ambiguity but prior to its disambiguation. Reading times for syntactically ambiguous sentences were compared to reading times for unambiguous controls in three self-paced reading experiments. All three kinds of constraints were found to be helpful, and when several constraints converged, ambiguity resolution was facilitated compared to when constraints conflicted. The importance of these constraint interactions for ambiguity resolution models is discussed.
When speakers refer to the same referent multiple times in a conversation, they tend to follow established patterns of usage, known as conversational precedents. Research has found that listeners expect speakers to follow precedents, and that this expectation guides their search for referents (Barr, D. J., & Keysar, B. (2002). Anchoring comprehension in linguistic precedents. Journal of Memory and Language, 46, 391–418). Recently, Metzing and Brennan (2003) (Metzing, C., & Brennan, S. E. (2003). When conceptual pacts are broken: partner-specific effects on the comprehension of referring expressions. Journal of Memory and Language, 49, 201–213) reported a speaker-specific effect for broken precedents that suggests early use of speaker information when precedents are broken. Results from two eyetracking experiments show that this speaker effect results from the late use of speaker information to recover from an early, partner-independent preemption effect. When a new description is heard, existing precedents preempt the mapping of the new description to an old referent. Later, listeners use speaker-information to inhibit precedents that are not known to the current speaker. Time-course data, as well as the results of a cognitive load manipulation, suggest that the preemption and speaker effects are supported by distinct processing systems. Our findings indicate that certain pragmatic effects in language comprehension are based on general expectations about language use, rather than assumptions about the beliefs and goals of particular speakers.
Acting together with others is a fundamental human ability. This raises the possibility that we take others' actions into account whenever somebody acts around us. Event-related fMRI was used to identify brain regions responsive to changes in cognitive processing when one and the same go-nogo task is performed alone or together with a co-actor performing a complementary task. Reaction times showed that participants integrated the potential action of their co-actor in their own action planning. Increased activation in ventral premotor cortex was found when participants acted upon stimuli referring to their own action alternative, but only when their partner performed a complementary task. This suggests that knowing about the potential actions of a partner increases the relevance of stimuli referring to oneself. Acting in the presence of a co-actor was also associated with increased orbitofrontal activation, indicating that participants monitored their performance more closely to make sure it really was their turn. These results suggest that our default mode is to interact with others.
The work reported here represents an attempt to extend the collaborative view of conversation (see, for example, Clark and Wilkes-Gibbs, 1986) to the level of the characteristics of individual words - in particular, word duration. Speaker dyads engaged in the tangram task which has been widely used for researching discourse-level behaviour. Four monologue and four dialogue trials were recorded from each speaker. Content words which appeared in at least seven trials were identified, and the successive tokens of these words were digitised and measured. Overall, words spoken in monologues were reliably longer than words spoken in dialogues, suggesting that speakers were conservative in their estimate of how intelligible a listener would find the words when no visual or linguistic feedback was available. In addition, monologue and dialogue tokens exhibited different shortening effects in response to repetition; in dialogue, speakers attenuated word durations on the second mention, but in monologue no shortening occurred between successive repetitions, although a gradual decrease in duration was observed across the four trials. These results parallel findings relating to referring expressions, where the decline in the number of words used to refer to an entity is much sharper in dialogue than in monologue. The results are discussed within the framework of the collaborative model.
People design what they say specifically for their conversational partners, and they adapt to their partners over the course of a conversation. A comparison of keyboard conversations involving a simulated computer partner (as in a natural language interface) with those involving a human partner (as in teleconferencing) yielded striking differences and some equally striking similarities. For instance, there were significantly fewer acknowl- edgments in human/computer dialogue than in human/human. However, regardless of the conversational partner, people expected connectedness across conversational turns. In ad- dition, the style of a partner's response shaped what people subsequently typed. These results suggest some issues that need to be addressed before a natural language computer interface will be able to hold up its end of a conversation.
The book from which these sections are excerpted (N. Chomsky, Rules and Representations, Columbia University Press, 1980) is concerned with the prospects for assimilating the study of human intelligence and its products to the natural sciences through the investigation of cognitive structures, understood as systems of rules and representations that can be regarded as “mental organs.” These mental structui′es serve as the vehicles for the exercise of various capacities. They develop in the mind on the basis of an innate endowment that permits the growth of rich and highly articulated structures along an intrinsically determined course under the triggering and partially shaping effect of experience, which fixes parameters in an intricate system of predetermined form. It is argued that the mind is modular in character, with a diversity of cognitive structures, each with its specific properties arid principles. Knowledge of language, of the behavior of objects, and much else crucially involves these mental structures, and is thus not characterizable in terms of capacities, dispositions, or practical abilities, nor is it necessarily grounded in experience in the standard sense of this term.Various types of knowledge and modes of knowledge acquisition are discussed in these terms. Some of the properties of the language faculty are investigated. The basic cognitive relation is “knowing a grammar”; knowledge of language is derivative and, correspondingly, raises further problems. Language as commonly understood is not a unitary phenomenon but involves a number of interacting systems: the “computational” system of grammar, which provides the representations of sound and meaning that permit a rich range of expressive potential, is distinct from a conceptual system with its own properties; knowledge of language must be distinguished from knowledge of how to use a language; and the various systems that enter into the knowledge and use of language must be further analyzed into their specific subcomponents.
What role does mutual knowledge play in the comprehension process? We compare two answers to this question for the comprehension of definite reference. The Restricted Search hypothesis assumes that addressees rely on the principle of optimal design and understand definite reference by restricting the search for referents to entities in common ground. The Unrestricted Search hypothesis assumes that the search for referents is not restricted to entities in common ground. Only the Unrestricted Search hypothesis predicts that entities that are not in common ground would interfere with comprehension of definite reference. Experiment 1 reveals such interference in increased errors and verification latencies during the resolution of pronouns. Experiment 2 demonstrates the interference by tracking the addressee's eye movements during the comprehension of demonstrative reference. We discuss alternative models of comprehension that could account for the results, and we describe the role that common ground plays in each model. We propose a Perspective Adjustment model that assumes a search for referents that is independent of common ground, coupled with a monitoring process that detects violations of common ground and adjusts the interpretation. This model assumes a role for common ground only when a correction is needed. We challenge both the assumption that addressees follow the principle of optimal design and the assumption that the principle is optimal.
I propose that research on the role of common ground in language use is often insensitive to considerations of parsimony. By ignoring parsimony, our experiments might fall short of addressing common ground and instead the resulting data support a simpler theory which is subsumed within the theory of common ground. In order to address this inherent conceptual confound, this paper proposes the subsuming theory criterion for experiments on common ground and mutual knowledge. In practical terms, to demonstrate an effect of common ground the design must keep common information constant and only vary whether or not it is common. The paper demonstrates how consistent use of this necessary criterion will allow us to make stronger claims about when common ground plays a role and when it does not To illustrate how experiments can be designed to satisfy the criterion, the paper evaluates some earlier studies which do and don't satisfy it In general, the study of language use could benefit from tightening the rules of evidence, allowing us to draw stronger conclusions about central concepts such as common ground.
Investigators of language understanding have made a number of idealizations in order to study it, but many of these idealizations have turned into dogmas—convictions that are impervious to evidence. Because of these dogmas, investigators have often ignored, dismissed, or ruled out of court common features of everyday language such as indirect meaning, word innovation, phrasal utterances, interjections, listener roles, listener background, specialized lexicons, joint actions by speakers and addressees, disfluencies, changes of mind, gestures, eye gaze, pretense, and quotations. I describe eleven common dogmas of understanding, some evidence against them, and some of the dangers they pose for the study of understanding. Using language is fundamentally social, I argue, and social features appear to influence understanding at many, perhaps most, levels of processing.
Past research has shown that when speakers refer to the same referent multiple times, they tend to standardize their descriptions by establishing linguistic precedents. In three experiments, we show that listeners reduce uncertainty in comprehension by taking advantage of these precedents. We tracked listeners' eye movements in a referential communication task and found that listeners identified referents more quickly when specific precedents existed than when there were none. Furthermore, we found that listeners expected speakers to adhere to precedents even in contexts where it would lead to referential overspecification. Finally, we provide evidence that the benefits of linguistic precedents are independent of mutual knowledge—listeners were not more likely to benefit from precedents when they were mutually known than when they were not. We conclude that listeners use precedents simply because they are available, not because they are mutually known.
Psycholinguistic models tend to depict speech production as a unidirectional process in which speakers create messages for generic addressees. Missing from such characterisations is an analysis of the process by which addressees influence the form and content of messages created for them. Addressees affect message form and content in at least two ways: (1) Their personal characteristics help define the common ground they share with the speaker and, hence, the knowledge the speaker can draw upon in formulating messages; (2) In conversation, their on-line responses serve to inform the speaker about changes in their knowledge state and, hence, about the communicative effectiveness of the messages transmitted. A series of experiments illustrating these processes is described.
When we search for a pen or a paper to jot down a note, the targets in these tasks are not specific visual patterns but rather categorically defined classes of objects. Although search is clearly guided by the visual features of specifically defined targets, it is not known whether search is similarly guided by the visual features defining an object class. We addressed this question by having subjects do two versions of an identical teddy-bear search task; one with a visually non-specific categorically defined target (e.g., “find the teddy-bear”) and the other with a specifically defined visual target (i.e., subjects were shown a target preview). Guidance was defined by the proportion of initial saccades directed to the target and by the cumulative probability of target fixation. As expected, we found evidence for target-specific guidance in both of these measures. More importantly, we also found evidence for guidance to categorically defined targets. Although categorical guidance was not as pronounced as target-specific guidance, subjects searching for the teddy-bear object class preferentially directed their initial saccades to teddy-bear targets and fixated these targets sooner than what would be expected from chance. In follow-up experiments we varied the repetition of specific targets and distractors in a categorical teddy-bear search task to determine how the availability of specific visual templates affects categorical search behavior. The data from these experiments suggest that subjects were forming non-specific visual templates for object classes and using these class-defining features to guide their categorical search.
A key issue in models of lexical production is whether a model of hearer knowledge influences speakers' productions. In this study we investigate whether the durational shortening of repeated words is a consequence of mutual knowledge, or if it is a byproduct of speaker internal processes. In a study of spontaneously produced narrations, in Experiment 1 we demonstrated that repeated words are shorter than their first mentions. In Experiment 2, we found that words are shortened more in the presence of a repeated hearer compared to a new hearer. Results from Experiment 3 demonstrate that shortening continues across narrations for repeated hearers, but is blocked by new hearers. These results indicate that a speaker's model of hearer knowledge affects production. Common Ground in Production: Effects of Mutual Knowledge on Word Duration Speakers commonly change the way they speak in order to conform to their interlocutors and other aspects of the speech context. These effects are pervasive and obvious; speakers raise their pitch when speaking with babies, code-switch only with appropriate interlocutors, and only use pronouns with referents that are accessible to others in the speech context. As noted by Clark and Wilkes-Gibbs (1986), in order for language to be a communicative process, there must be some sort of shared knowledge and experience between interlocutors. The knowledge that is shared between interlocutors is often referred to as common ground, "the background, the context, for everything the participants both do and say" (Clark, 1994, p. 990). While common ground is information that both speaker and hearer share, mutual knowledge is what the speaker, or hearer, knows (or assumes) about what her conversation partner knows (see Cark & Marshall, 1981). For example, ...
In Gerrig, Ohaeri, and Brennan (this issue) we called into question whether "illusory transparency" (Keysar, 1994) is actually an interesting and perplexing phenomenon, or whether readers are simply perplexed when protagonists in stories behave uncooperatively or irrationally. In this commentary, we highlight data from our original experiments that were not addressed by Keysar (this issue). We suggest that serious demand characteristics, rather than illusory transparency, explain participants' greatly varying judgments across experiments. We also reiterate our concern that, with respect to expectations of cooperative and rational behavior, the speakers in Keysar's scenarios are behaving quite badly. It is questionable whether important theoretical generalizations can or should be drawn from this paradigm.
discuss sentence comprehension, focusing on the relation between language structure and language processing at this level of analysis / presents a history of sentence comprehension over the past 35 yrs that reviews now classic experiments and theories that drove these research efforts / summarize recent experimental efforts and the evolving theories and metatheories that have driven these experimental investigations; this discussion includes a consideration of the important evolving notions of the lexical/syntactic interface (PsycINFO Database Record (c) 2012 APA, all rights reserved)
Three speakers read aloud meaningful grammatical English sentences at a fast rate. Some of these sentences contained common maxims and stereotyped phrases. Other sentences which were less familiar contained, in similar phonetic environments, certain test words that also occurred in the stereotyped sentences. The test words were “excised”, i.e., gated out of all the sentences, and listening tests were performed by 43 listeners. Quite apart from these listening tests two operational measures of redundancy were derived. One measure was based on a Markovian model of perception and readers were asked to guess what word n would be after they had read all the words through word n–1 in a sentence. The second measure was based on the hypothesis that people reserve their final decision on the recognition of each word until they perceive an entire sentence. Readers read an entire sentence in the midst of which a word was represented by a dash. They were then asked to guess what the missing word was. Two groups of thirty readers each performed these written tests on the sentences from which the test words had been excised. Indexes of redundancy were computed for each test word from the percent of correct guesses that occurred for the two written tests.
Pairs of friends wrote descriptions of abstract stimuli for one another. Later, each subject attempted to identify the referents of three sets of messages: those created by the subject, those created by the friend, and those created by a stranger for his J her own friend. Subjects were most accurate with their own messages, but, more interestingly, they were more accurate with their friends' than with strangers' messages. The lexical characteristics and content of messages for friends were compared with those intended for a generic ‘other student’ or for oneself in Fussell and Krauss (1989). Contrary to our expectations, friends' messages were similar to those for another student in length, vocabulary distribution, and figurative language use, while both differed significantly from messages for oneself: The findings are discussed within a ‘common ground’ framework for communication.
Early research on spoken language established that words extracted from their original sentential contexts were often very hard to understand. Their intelligibility was modulated by the context from which they were taken: Words excised from highly predictive contexts were less intelligible than ones taken from less predictive contexts, suggesting that talkers compensate for weak context by improving articulation. Curiously, the only study to use children as talkers in this paradigm found the reverse pattern—better intelligibility of words from predictive contexts. Experiment 1 of the current study replicated this odd result, and showed that it only occurred when children were reading prepared materials; the “normal” adult-like pattern was found in the children's spontaneous conversation. Experiment 2 demonstrated that adults can be induced to produce the “reverse” pattern, when they are placed in a challenging reading situation. Experiment 3 showed that adults with high verbal skills produce the normal adult-like pattern when reading aloud, whereas adults with only average verbal skills do not. Collectively, the results indicate that the traditional effect of redundancy can be found only when talkers have good control of the production situation. Children as young as ten can in fact compensate for contextual weakness through articulatory enhancement, but only in conversational situations that afford such control.
Speakers in conversation routinely engage in audience design. That is, they construct their utterances to be understood by particular addressees. Standard accounts of audi-ence design have frequently appealed to the notion of common ground. On this view, speakers produce well-designed utterances by expressly considering the knowledge they take as shared with addressees. This article suggests that conversational com-mon ground, rather than being a category of specialized mental representations, is more usefully conceptualized as an emergent property of ordinary memory pro-cesses. This article examines 2 separate but equally important processes: commonal-ity assessment and message formation. Commonality assessment involves the re-trieval of memory traces concerning what information is shared with an addressee, whereas message formation involves deciding how to use that information in conver-sation. Evidence from the CallHome English corpus of telephone conversations shows how each of these processes is rooted in basic aspects of human memory. The overall goal of this article is to demonstrate the need for a more cognitive psychologi-cal account of conversational common ground. Consider this excerpt from a conversation between two friends who have not spo-ken with each other for some time: (1) A: Oh first of all I have Shana's shower coming up that I have to do. B: Ah, that's right. A: That's going to be like a huge like three day effort with all the cooking and cleaning and like actually party [sic] that I have to do. B: Is there anyone you can get to help you? A: Um Jessica's going to help and Beth might because you see, Diane is here now. B: Oh okay. [#4913, 440.30] DISCOURSE PROCESSES, 40(1), 1–35 Copyright © 2005, Lawrence Erlbaum Associates, Inc.
Perspective-taking is central to much social interaction, but the processes by which it is accomplished are poorly understood. The current study examines accuracy and bias in one type of perspective-taking: inferences about what others know. Twenty-two New York City landmarks were presented in three conditions: Picture Only, Picture-+ Name and Name Only. Subjects estimated the proportion of short- and long-term New York City residents who could identify each landmark from its picture. They also rated their subjective recognition of the stimuli. Subjects in all three conditions were good at estimating stimulus identifiability, but their estimates were biased in the direction of their own knowledge. Estimates of the difference in identifications by short-and long-term residents were relatively inaccurate, probably because the two groups differed less than anticipated. For most but not all subjects, subjective feelings of recognition were significantly correlated with estimates of identifiability. We conclude that perceptions of the distribution of knowledge are socially shared.
Drawing on recent findings in the cognitive and neurosciences, this article discusses how people manage to predict each other’s actions, which is fundamental for joint action. We explore how a common coding of perceived and performed actions may allow actors to predict the what, when, and where of others’ actions. The “what” aspect refers to predictions about the kind of action the other will perform and to the intention that drives the action. The “when” aspect is critical for all joint actions requiring close temporal coordination. The “where” aspect is important for the online coordination of actions because actors need to effectively distribute a common space. We argue that although common coding of perceived and performed actions alone is not sufficient to enable one to engage in joint action, it provides a representational platform for integrating the actions of self and other. The final part of the paper considers links between lower-level processes like action simulation and higher-level processes like verbal communication and mental state attribution that have previously been at the focus of joint action research.
In conversation speakers design their utterances to be understood against the common ground they share with their addressees—their common experience, expertise, dialect, and culture. That ordinarily gives addressees an advantage over overhearers in understanding. Addressees have an additional advantage, we propose, because they can actively collaborate with speakers in reaching the mutual belief that they have understood what was said, whereas overhearers cannot. As evidence for the proposal, we looked at triples of people in which one person told another person in conversation how to arrange 12 complex figures while an overhearer tried to arrange them too. All three began as strangers with the same background information. As predicted, addressees were more accurate at arranging the figures than overhearers even when the overhearers heard every word. Other evidence suggests that the very process of understanding is different for addressees and overhearers.
This study investigated whether production is adapted to comprehension. Experiment 1 examined whether or not instruments are explicitly mentioned in accordance with their likelihood of being inferred. Subjects read and retold 20 short stories. An analysis of the transcribed stories revealed that atypical instruments (e.g., an ice pick in a stabbing) were specified significantly more often than typical instruments (e.g., a knife in a stabbing), and instruments that were important in the story were specified slightly more often than unimportant instruments. Experiment 2 examined the mechanism accounting for this phenomenon and revealed that the speaker's beliefs about the listener's knowledge of the instrument had little impact on whether and how the instrument was mentioned. A quantitative model for mentioning instruments was proposed. Instrument concepts can be selected at three points in the production process: (1) during creation of a macroproposition encoding the main idea of a discourse episode, (2) during the elaboration of the macroproposition, and (3) during the monitoring and repair process.
In two experiments, we explored the time course and flexibility with which speakers’ eye gaze can be used to disambiguate referring expressions in spontaneous dialog. Naive director/matcher pairs were separated by a barrier and saw each other’s faces but not their displays. Displays held identical objects, with the matcher’s arranged in a row and the director’s mirroring the matcher’s or else in a circle (Experiment 1) or in a reversed row (Experiment 2). Directors instructed matchers to move targets, which were unique or had a competitor nearby or far away. When mirrored displays held far competitors, matchers used directors’ eye gaze to identify targets before the linguistic point of disambiguation. Reversed displays caused substantial competition, yet matchers still identified targets before the linguistic point of disambiguation, showing an ability to rapidly re-map directors’ eye gaze. Our findings indicate eye gaze is a powerful and flexible disambiguating cue in referential communication.
In this paper, we develop an account of the types of experiences through which speakers learn to design their utterances for particular addressees. We argue that there are two important aspects of conversational situations relevant to considerations of audience design. First, speakers must become aware that audience design is necessary in the current setting. Second, they must frequently overcome other tendencies toward consistency and brevity of expression. To assess the impact of both of these factors, we conducted a referential communication experiment in which Directors described arrays of picture cards for two independent Matchers. In the early rounds, both Matchers were present and each possessed a different subset of the Directors’ cards. In later rounds, only one of the two Matchers was present at a time and worked with the entire set of cards. We evaluated the degree to which Directors’ descriptions showed evidence of audience design by focusing on critical rounds when the Directors described cards that the current Matcher had not previously shared. Directors generally appeared sensitive to the distinction between shared and nonshared items. Additionally, there was more evidence of adjustment at the second partner change, suggesting that the Directors had learned something about the kinds of descriptions required in this situation. Our results suggest that it is important to consider the nature of speakers’ experiences of interacting in a particular situation when making claims about the presence or absence of audience design.
When two people in conversation refer repeatedly to objects, they typically converge on the same (or similar) referring expressions. The repeated use of expressions by people in the same conversation has been called lexical entrainment. Lexical entrainment may emerge from the precedent of associating objects with expressions (and the perspectives they encode), or else from achieving conceptual pacts, or temporary, flexible agreements to view an object in a particular way (in which case the precedent is encoded as specific to a particular partner). We had people interact with a confederate speaker, entraining on shared perspectives (e.g., “the shiny cylinder”) during repeated references to objects. Then either the original speaker or a new speaker used either the original expression or a new one (“the silver pipe”) to refer to the previously discussed object. Upon hearing the original expressions, addressees looked at and then touched the target objects equally quickly regardless of speaker. However, with new expressions, there was partner-specific interference: addressees were slower to look at the object when the new expression was uttered by the original speaker than when the new expression was uttered by the new speaker. This suggests that the representations in memory from which entrainment emerges do encode a partner-specific cue, leading addressees to expect that a speaker should continue to use an entrained-upon expression unless a contrast in meaning is implicated. There appears to be no such interference when a new partner uses a new expression.
The problems of access—retrieving linguistic structure from some mental grammar —and disambiguation—choosing among these structures to correctly parse ambiguous linguistic input—are fundamental to language understanding. The literature abounds with psychological results on lexical access, the access of idioms, syntactic rule access, parsing preferences, syntactic disambiguation, and the processing of garden-path sentences. Unfortunately, it has been difficult to combine models which account for these results to build a general, uniform model of access and disambiguation at the lexical, idiomatic, and syntactic levels. For example, psycholinguistic theories of lexical access and idiom access and parsing theories of syntactic rule access have almost no commonality in methodology or coverage of psycholinguistic data. This article presents a single probabilistic algorithm which models both the access and disambiguation of linguistic knowledge. The algorithm is based on a parallel parser which ranks constructions for access, and interpretations for disambiguation, by their conditional probability. Low-ranked constructions and interpretations are pruned through beam-search; this pruning accounts, among other things, for the garden-path effect. I show that this motivated probabilistic treatment accounts for a wide variety of psycholinguistic results, arguing for a more uniform representation of linguistic knowledge and for the use of probabilistically-enriched grammars and interpreters as models of human knowledge of and processing of language.
An experiment examines talkers' utterances of words produced for the first time in a monolog (“new” words) or for the second time (“old” words). The finding is that talkers distinguish old words by shortening them. Two experiments show that old words are less intelligible than new words presented in isolation, but probably are not less identifiable in context. We infer that talkers may attenuate their productions of words when they can do so without sacrificing communicative efficacy. Old words can be reduced because they are repetitions of earlier presented items and because of the contextual support they receive. Two final experiments show that listeners can identify new and old words as such and that they can use information that a word is old more or less as they would use an anaphor to promote retrieval of the earlier production in its context.
In order to investigate whether addressees can make immediate use of speaker-based constraints during reference resolution, participant addressees’ eye movements were monitored as they helped a confederate cook follow a recipe. Objects were located in the helper’s area, which the cook could not reach, and the cook’s area, which both could reach. Critical referring expressions matched one object (helper’s area) or two objects (helper’s and cook’s areas), and were produced when the cook’s hands were empty or full, which defined the cook’s reaching ability constraints. Helper’s first and total fixations showed that they restricted their domain of interpretation to their own objects when the cook’s hands were empty, and widened it to include the cook’s objects only when the cook’s hands were full. These results demonstrate that addressees can quickly take into account task-relevant constraints to restrict their referential domain to referents that are plausible given the speaker’s goals and constraints.
Addressees’ eye movements were tracked as they followed instructions given by a confederate speaker hidden from view. Experiment 1 used objects in common ground (known to both participants) or privileged ground (known to the addressee). Although privileged objects interfered with reference to an identical object in common ground, addressees were always more likely to look at an object in common ground than privileged ground. Experiment 2 used definite and indefinite referring expressions with early or late points of disambiguation, depending on the uniqueness of the display objects. The speaker’s and addressee’s perspectives matched when the speaker was accurately informed about the display, and mismatched when the speaker was misinformed. When perspectives matched, addressees identified the target faster with early than with late disambiguation displays. When perspectives mismatched, addressees still identified the target quickly, showing an ability to use the speaker’s perspective. These experiments demonstrate that although addressees cannot completely ignore information in privileged ground, common ground and perspective each have immediate effects on reference resolution.
When we think of the ways we use language, we think of face-to-face conversations, telephone conversations, reading and writing, and even talking to oneself. These are arenas of language useâtheaters of action in which people do things with language. But what exactly are they doing with language? What are their goals and intentions? By what processes do they achieve these goals? In these twelve essays, Herbert H. Clark and his colleagues discuss the collective nature of languageâthe ways in which people coordinate with each other to determine the meaning of what they say. According to Clark, in order for one person to understand another, there must be a "common ground" of knowledge between them. He shows how people infer this "common ground" from their past conversations, their immediate surroundings, and their shared cultural background. Clark also discusses the means by which speakers design their utterances for particular audiences and coordinate their use of language with other participants in a language arena. He argues that language use in conversation is a collaborative process, where speaker and listener work together to establish that the listener understands the speaker's meaning. Since people often use words to mean something quite different from the dictionary definitions of those words, Clark offers a realistic perspective on how speakers and listeners coordinate on the meanings of words. This collection presents outstanding examples of Clark's pioneering work on the pragmatics of language use and it will interest psychologists, linguists, computer scientists, and philosophers.
Thesis (Ph. D.)--Stanford University, 1990. Submitted to the Department of Psychology. Copyright by the author.