Two experiments examined the restriction of referential domains during unscripted conversation by analyzing the modification and online interpretation of referring expressions. Experiment 1 demonstrated that from the earliest moments of processing, addressees interpreted referring expressions with respect to referential domains constrained by the conversation. Analysis of eye movements during the conversation showed elimination of standard competition effects seen with scripted language. Results from Experiment 2 pinpointed two pragmatic factors responsible for restriction of the referential domains used by speakers to design (...) referential expressions and demonstrated that the same factors predict whether addressees consider local competitors to be potential referents during online interpretation of the same expressions. These experiments demonstrate, for the first time, that online interpretation of referring expressions in conversation is facilitated by referential domains constrained by pragmatic factors that predict when addressees are likely to encounter temporary ambiguity in language processing. (shrink)

Upon hearing a scalar adjective in a definite referring expression such as “the big…,” listeners typically make anticipatory eye movements to an item in a contrast set, such as a big glass in the context of a smaller glass. Recent studies have suggested that this rapid, contrastive interpretation of scalar adjectives is malleable and calibrated to the speaker's pragmatic competence. In a series of eye‐tracking experiments, we explore the nature of the evidence necessary for the modulation of pragmatic inferences in (...) language comprehension, focusing on the complementary roles of top–down information ‐ (knowledge about the particular speaker's pragmatic competence) and bottom‐up cues (distributional information about the use of scalar adjectives in the environment). We find that bottom‐up evidence alone (e.g., the speaker says “the big dog” in a context with one dog), in large quantities, can be sufficient to trigger modulation of the listener's contrastive inferences, with or without top‐down cues to support this adaptation. Further, these findings suggest that listeners track and flexibly combine multiple sources of information in service of efficient pragmatic communication. (shrink)

During communication, we form assumptions about what our communication partners know and believe. Information that is mutually known between the discourse partners—their common ground—serves as a backdrop for successful communication. Here we present an introduction to the focus of this topic, which is the role of memory in common ground and language use. Two types of questions emerge as central to understanding the relationship between memory and common ground, specifically questions having to do with the representation of common ground in (...) memory, and the use of common ground during language processing. (shrink)

How do speakers design what they say in order to communicate effectively with groups of addressees who vary in their background knowledge of the topic at hand? Prior findings indicate that when a speaker addresses a pair of listeners with discrepant knowledge, that speakers Aim Low, designing their utterances for the least knowledgeable of the two addressees. Here, we test the hypothesis that speakers will depart from an Aim Low approach in order to efficiently communicate with larger groups of interacting (...) partners. Further, we ask whether the cognitive demands of tracking multiple conversational partners' perspectives places limitations on successful audience design. We find that speakers can successfully track information about what up to four of their partners do and do not know in conversation. When addressing groups of 3–4 addressees at once, speakers design language based on the combined knowledge of the group. These findings point to an audience design process that simultaneously represents the perspectives of multiple other individuals and combines these representations in order to design utterances that strike a balance between the different needs of the individuals within the group. (shrink)

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. (shrink)

Communicating with multiple addressees poses a problem for speakers: Each addressee necessarily comes to the conversation with a different perspective—different knowledge, different beliefs, and a distinct physical context. Despite the ubiquity of multiparty conversation in everyday life, little is known about the processes by which speakers design language in multiparty conversation. While prior evidence demonstrates that speakers design utterances to accommodate addressee knowledge in multiparty conversation, it is unknown if and how speakers encode and combine different types of perspective information. (...) Here we test whether speakers encode the perspective of multiple addressees, and then simultaneously consider their knowledge and physical context during referential design in a three‐party conversation. Analyses of referential form—expression length, disfluency, and elaboration rate—in an interactive multiparty conversation demonstrate that speakers do take into consideration both addressee knowledge and physical context when designing utterances, consistent with a knowledge‐scene integration view. These findings point to an audience design process that takes as input multiple types of representations about the perspectives of multiple addressees, and that bases the informational content of the to‐be‐designed utterance on a combination of the perspectives of the intended addressees. (shrink)

Written memoranda of conversations, or memcons, provide a near‐contemporaneous record of what was said in conversation, and offer important insights into the activities of high‐profile individuals. We assess the impact of writing a memcon on memory for conversation. Pairs of participants engaged in conversation and were asked to recall the contents of that conversation 1 week later. One participant in each pair memorialized the content of the interaction in a memcon shortly after the conversation. Participants who generated memcons recalled more (...) details of the conversations than participants who did not, but the content of recall was equally and largely accurate for both participants. Remarkably, only 4.7% of the details of the conversation were recalled by both of the partners after a week delay. Contemporaneous note‐taking appears to enhance memory for conversation by increasing the amount of information remembered but not the accuracy of that information. These findings have implications for evaluating the testimony of participants on conversations with major political or legal ramifications. (shrink)

Semantic memory encompasses one's knowledge about the world. Distributional semantic models, which construct vector spaces with embedded words, are a proposed framework for understanding the representational structure of human semantic knowledge. Unlike some classic semantic models, distributional semantic models lack a mechanism for specifying the properties of concepts, which raises questions regarding their utility for a general theory of semantic knowledge. Here, we develop a computational model of a binary semantic classification task, in which participants judged target words for the (...) referent's size or animacy. We created a family of models, evaluating multiple distributional semantic models, and mechanisms for performing the classification. The most successful model constructed two composite representations for each extreme of the decision axis (e.g., one averaging together representations of characteristically big things and another of characteristically small things). Next, the target item was compared to each composite representation, allowing the model to classify more than 1,500 words with human‐range performance and to predict response times. We propose that when making a decision on a binary semantic classification task, humans use task prompts to retrieve instances representative of the extremes on that semantic dimension and compare the probe to those instances. This proposal is consistent with the principles of the instance theory of semantic memory. (shrink)

We agree with Pickering & Garrod's (P&G's) proposal that dialogue is an important empirical and theoretical test bed for models of language processing. However, we offer two cautionary notes. First, the enterprise will require explicit computational models. Second, such models will need to incorporate both joint and separate speaker and hearer commitments in ways that go beyond priming and alignment.

Speakers tend to produce disfluencies when naming unexpected or complex items; in turn, when perceiving disfluency, listeners tend to expect upcoming reference to items that are unexpected or complex to name. In two experiments, we examined if these disfluency-based expectations are routine, or instead, if they adapt to the way the speaker uses disfluency in the current context in a talker-specific manner. Participants listened to instructions to look at objects in contexts with several images, some of which lacked conventional names. (...) We manipulated the co-occurrence of disfluency and reference to novel versus familiar objects in a single talker situation (Experiment 1) and in a multi-talker situation (Experiment 2). In the predictive condition, disfluent expressions referred to novel objects, and fluent expressions referred to familiar objects. In the nonpredictive condition, fluent and disfluent trials referred to either familiar or novel objects. Participants’ gaze revealed that listeners more readily predicted familiar images for fluent trials and novel images for disfluent trials in the predictive condition than in the nonpredictive condition. In sum, listeners adapted their expectations about upcoming words based on recent experience with disfluency. Disfluency is not invariably processed, but instead a cue that is flexibly interpreted depending on the local context even in a multi-talker setting. (shrink)

The way we refer to things in the world is shaped by the immediate physical context as well as the discourse history. But what part of the discourse history is relevant to language use in the present? In four experiments, we combine the study of task‐based conversation with measures of recognition memory to examine the role of physical contextual cues that shape what speakers perceive to be a part of the relevant discourse history. Our studies leverage the differentiation effect, a (...) phenomenon in which speakers are more likely to use a modified expression to refer to an object (e.g., dotted sock) if they had previously described a similar object (e.g., striped sock) than when they had not described a similar object. Two physical cues—the background that framed the to‐be‐described pictures and the position of the pictures in the display—were manipulated to alter perceptions about the relevant discourse context. We measured the rate with which speakers modify referring expressions to differentiate current from past referents. Recognition memory measures following the conversation probed what was and was not remembered about past discourse referents and contexts. Analysis of modification rates indicated that these contextual grouping cues shaped perceptions about the relevant discourse context. The contextual cues did not affect memory for the referents, but the memory for past referents was better for speakers than for listeners. Our findings show that perceptions about the relevant discourse history are a key determinant of how language is used in the moment but also that conversational partners form asymmetric representations of the discourse history. (shrink)

We agree with Pickering & Garrod's (P&G's) claim that theories of language processing must address the interconnection of language production and comprehension. However, we have two concerns: First, the central notion of context when predicting what another person will say is underspecified. Second, it is not clear that P&G's dual-mechanism model captures the data better than a single-mechanism model would.

Speech disfluencies can signal that a speaker is about to refer to something difficult to name. In two experiments, we found evidence that 4-year-olds, like adults, flexibly interpret a particular partner’s disfluency based on their estimate of that partner’s knowledge, derived from the preceding conversation. In entrainment trials, children established partner-specific shared knowledge of names for tangram pictures with one or two adult interlocutors. In each test trial, an adult named one of two visible tangrams either fluently or disfluently while (...) children’s eye-movements were monitored. We manipulated speaker knowledge in the test trials. In Experiment 1, the test-trial speaker was the same speaker from entrainment or a naïve experimenter; in Experiment 2, the test-trial speaker had been one of the child’s partners in entrainment and had seen half of the tangrams. When hearing disfluent expressions, children looked more at a tangram that was unfamiliar from the speaker’s perspective; this systematic disfluency effect disappeared in Experiment 1 when the speaker was entirely naïve, and depended on each speaker’s entrainment experience in Experiment 2. These findings show that 4-year-olds can keep track of two different partners’ knowledge states, and use this information to determine what should be difficult for a particular partner to name, doing so efficiently enough to guide online interpretation of disfluent speech. (shrink)