Cue effectiveness in communicatively efficient discourse production

Recent years have seen a surge in accounts motivated by information theory that consider language production to be partially driven by a preference for communicative efficiency. Evidence from discourse production (i.e., production beyond the sentence level) has been argued to suggest that speakers distribute information across discourse so as to hold the conditional per-word entropy associated with each word constant, which would facilitate efficient information transfer (Genzel & Charniak, 2002). This hypothesis implies that the conditional (contextualized) probabilities of linguistic units affect speakers' preferences during production. Here, we extend this work in two ways. First, we explore how preceding cues are integrated into contextualized probabilities, a question which so far has received little to no attention. Specifically, we investigate how a cue's maximal informativity about upcoming words (the cue's effectiveness) decays as a function of the cue's recency. Based on properties of linguistic discourses as well as properties of human memory, we analytically derive a model of cue effectiveness decay and evaluate it against cross-linguistic data from 12 languages. Second, we relate the information theoretic accounts of discourse production to well-established mechanistic (activation-based) accounts: We relate contextualized probability distributions over words to their relative activation in a lexical network given preceding discourse.     

Language context guides memory content

The relationship between language and memory was examined by testing accessibility of general knowledge across two languages in bilinguals. Mandarin-English speakers were asked questions such as “name a statue of someone standing with a raised arm while looking into the distance” and were more likely to name the Statue of Liberty when asked in English and the Statue of Mao when asked in Mandarin. Multivalent information (i.e., multiple possible answers to a question) and bivalent information (i.e., two possible answers to a question) were more susceptible to language dependency than univalent information (i.e., one possible answer to a question). Accuracy of retrieval showed language-dependent memory effects in both languages, while speed of retrieval showed language-dependent memory effects only in bilinguals’ more proficient language. These findings suggest that memory and language are tightly connected and that linguistic context at the time of learning may become integrated into memory content.     

Issue-Event Extensions and Interaction Involvement Text-Based and Meaning-Based Discourse Strategies

Prior research focusing on textual features in the talk of high and low-involved speakers has characterized the discourse strategy of low-involved speakers as text-based and that of high-involved speakers as meaning-based. This study examined the extensions of issue-event structures by high-involved and low-involved participants to determine if this characterization of discourse strategies held on a pragmatic level as well. Results indicated that low-involved communicators extended the event significantly more directly than high-involved communicators. Furthermore, the issue extensions of low-involved communicators were shown to be more sensitive to the comprehensibitity and information value of the stimulus utterance than the issue extensions of high-involved communicators. The results confirm that low-involved communicators have more difficulty than high-involved communicators in responding to the flow of conversation and that low-involved communicators compensate by relying on a text-based discourse strategy.     

Referential Form and Memory for the Discourse History

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.     

Turn-taking and speech act patterns in the discourse of senile dementia of the Alzheimer's type patients

Conversational discourse patterns of 11 normal elderly and 11 senile dementia of the Alzheimer's type (SDAT) patients engaged in dyadic interaction with an examiner were examined. Differences in word usage, turn taking, and speech act production were investigated both for the two-subject groups and for the examiner's conversations with each group. Compensatory shifts in discourse by participants are identified. For the subject, differences were shown on words per turn with SDAT subjects speaking in shorter turns and in nonverbal responses with SDAT subjects using this strategy more frequently. Speech act categories of Requestives and Assertives also differed with SDAT subjects using more Requestives and fewer Assertives. The SDAT subjects had significantly more occurrences of unintelligible utterances. For the examiner, words per turn differed with the examiner using shorter turns with SDAT subjects. No differences were shown in the examiner's patterns of speech act usage, nonverbal responses, or intelligibility. In general, these results indicate significant discourse differences in the words per turn level for all participants and speech act levels of conversation for SDAT subjects. They also indicate generally maintained interaction patterns by speakers so that the discourse genre of conversation is sustained. The pattern of compensatory shifts in discourse suggests retained flexibility in the communication system of early and mid stage SDAT patients.     

Adult-child discourse: Developmental interaction between information processing and linguistic knowledge

In the present study, the discourse interaction between adult and child was examined in terms of the content of their utterances, and the linguistic and contextual relations between their messages, in order to investigate how children use the information from adults' input sentences to form contingent responses. The analyses described were based on longitudinal data from four children from approximately 21 to 36 months of age. Categories of child discourse, their development and their interactions with aspects of prior adult utterances form the major results of the study. Child utterances were identified as adjacent (immediately preceded by an adult utterance), or as nonadjacent (not immediately preceded by an adult utterance). Adjacent utterances were either contingent (shared the same topic and added new information relative to the topic of the prior utterance), imitative (shared the same topic but did not add new information), or noncontingent (did not share the same topic). From the beginning, the adjacent speech was greater than nonadjacent speech. Contingent speech increased over time; in particular, linguistically contingent speech (speech that expanded the verb relation of the prior adult utterance with added or replaced constituents within a clause) showed the greatest developmental increase. Linguistically contingent speech occurred more often after questions than nonquestions. The results are discussed in terms of how the differential requirements for processing information in antecedent messages is related to language learning.     

Studies in dialogue and discourse. III. Utterance structure and utterance function in interrogative sequences

Natural conversations in first-grade classrooms are analyzed. Exchanges that are initiated and extended by questioning are the focus of attention, and analyses are directed principally to a comparison of adult-initiated and child-initiated interrogative series. Differences in structural features are found for each component utterance of an interrogative unit of discourse, namely, question, response, and confirmation. Adults ask more constraining questions than children, and children respond in more complex although less appropriate ways to children's questions than to adults' questions. The type of discourse within which an utterance is located influences the features of children's responses to children's but not to adults' questions. These differences are interpreted as reflecting differences in the relative power of adults and children in these settings.This work has been supported by a grant from the National Science Foundation for research on Social Effects on Language and Communication, Grant No. GS-3001.     

Representing logical and semantic structure of knowledge acquired from discourse

A network model of logical and semantic structures from which speakers or writers generate linguistic messages at the discourse level is presented. While linguistic structures were considered in developing the model, the semantic and logical networks are defined without reference to linguistic structures and thus may be used to represent knowledge structures acquired from both linguistic and nonlinguistic sources. A second problem addressed is that of determining what logical and semantic information is acquired when a text is understood. To assess acquired knowledge, a procedure is presented for coding a subject's verbal reconstruction of knowledge acquired from a presented text (or other input) against the logical and semantic structure from which the text (or other input) was derived. The procedures are illustrated using data obtained from children who were asked to “retell” simple narrative stories.     

Rules of Meaning and Practical Reasoning

Can there be rules of language which serve both to determine meaning and to guide speakers in ordinary linguistic usage, i.e., in the production of speech acts? We argue that the answer is no. We take the guiding function of rules to be the function of serving as reasons for actions, and the question of guidance is then considered within the framework of practical reasoning. It turns out that those rules that can serve as reasons for linguistic utterances cannot be considered as normative or meaning determining. Acceptance of such a rule is simply equivalent to a belief about meaning, and does not even presuppose that meaning is determined by rules. Rules that can determine meaning, on the other hand, i.e., rules that can be regarded as constitutive of meaning, are not capable of guiding speakers in the ordinary performance of speech acts.     

Switching pauses in adult-adult and child-child turn takings: An initial study

Based on the Japanese conversation obtained from five adult-adult and five child-child dyads, the present study investigated whether the duration of switching pause (silence between two different speakers) differed depending on the existence of linguistic cue (e.g., anaphoric reference) and/or the phonetic cue (rising intonation). Switching pause durations were measured acoustically whereas the phonetic cue was examined by a listening test. The results showed that switching pause durations are likely to be short when they are accompanied by the linguistic cue whereas they are likely to be long when accompanied by no cue. This tendency was observed both in adult and child dyads. The results also showed that the ratios of using the linguistic and/or the phonetic cue are almost the same for adults and children.     

Cognitive strategies in discourse processing: A comparison of Chinese and English speakers

Three experiments tested the hypothesis that, when processing a normative language, individuals tend to adopt the discourse processing strategies that they use in their native language. In a natural language processing task, native speakers of English and Chinese read English passages that were heavily loaded with zero anaphora. Native Chinese speakers found the passages more comprehensible than did native English speakers, presumably because zero anaphora occurs much more often in Chinese than in English. However, in an artificial laboratory task, native Chinese speakers did not find it easier to recover the missing referents than did native English speakers, suggesting that the two groups were equally capable of reference retrieval although they differed in their natural language discourse-processing strategies. These results suggest that the strategies used for processing foreign-language discourse are influenced by those used to process native-language discourse.This research was supported in part by Dean's Small Grants from the Graduate School of the University of Colorado to Liang Tao and United States Army Research Institute Contracts MDA903-90-K-0066 and MDA903-93-K-0010 to the University of Colorado (Alice Healy, Principal Investigator).We thank Barbara Fox, Lise Menn, and Nanfang Jiang for stimulating discussions about this research.An earlier version of this paper was presented at the First International Conference on Chinese Linguistics held in Singapore, 1992.     

Question form analysis in discourse agreement between a mother and a language-learning child

This study analyzed the question form used in discourse between a mother and her language-learning child. Spontaneous taped data of mother-child interactions were collected at three different time periods. Initially the child was 2;3 and data collection stopped at 2;5. Analysis of dialogue units based on protocols of discourse proposed by Mishler (1974) was conducted. Results indicated that question chaining occurred most frequently in discourse, followed by arching, then embedding. All question forms except one were initiated by the mother. Evidence also indicated that the subject did not have the linguistic competence of allwh-question forms by 2;5. This study indicated that question chaining was the most frequent type of discourse analysis with this dyad.     

Producing Pronouns and Definite Noun Phrases: Do Speakers Use the Addressee’s Discourse Model?

We report two experiments that investigated the widely held assumption that speakers use the addressee's discourse model when choosing referring expressions (e.g., Ariel, 1990; Chafe, 1994; Givón, 1983; Prince, 1985), by manipulating whether the addressee could hear the immediately preceding linguistic context. Experiment 1 showed that speakers increased pronoun use (and decreased noun phrase use) when the referent was mentioned in the immediately preceding sentence compared to when it was not, even though the addressee did not hear the preceding sentence, indicating that speakers used their own, privileged discourse model when choosing referring expressions. The same pattern of results was found in Experiment 2. Speakers produced more pronouns when the immediately preceding sentence mentioned the referent than when it mentioned a referential competitor, regardless of whether the sentence was shared with their addressee. Thus, we conclude that choice of referring expression is determined by the referent's accessibility in the speaker's own discourse model rather than the addressee's.     

The effect of question expectedness and experience on lying about intentions

In recent years researchers have started to focus on lying about intentions (Granhag, 2010). In the present experiment participants were interviewed about their forthcoming trip. We tested the hypothesis that liars (N=43) compared to truth tellers (N=43) would give fewer details to unexpected questions about planning, transportation and the core event, but an equal amount or more detail to expected questions about the purpose of the trip. We also tested the hypothesis that participants who had previously experienced the intention (i.e., they had made such a trip before) would give more detail than those who had never experienced the intended action. The unexpected question hypothesis was supported, whereas the previous experience effect only emerged in interactions. The benefit of using different types of questions for lie detection purposes is discussed.     

What we know about what we have never heard: evidence from perceptual illusions

Are speakers equipped with preferences concerning grammatical structures that are absent in their language? We examine this question by investigating the sensitivity of English speakers to the sonority of onset clusters. Linguistic research suggests that certain onset clusters are universally preferred (e.g., bd>lb). We demonstrate that such preferences modulate the perception of unattested onsets by English speakers: Monosyllabic auditory nonwords with onsets that are universally dispreferred (e.g., lbif) are more likely to be classified as disyllabic and misperceived as identical to their disyllabic counterparts (e.g., lebif) compared to onsets that are relatively preferred across languages (e.g., bdif). Consequently, dispreferred onsets benefit from priming by their epenthetic counterpart (e.g., lebif-lbif) as much as they benefit from identity priming (e.g., lbif-lbif). A similar pattern of misperception (e.g., lbif-->lebif) was observed among speakers of Russian, where clusters of this type occur. But unlike English speakers, Russian speakers perceived these clusters accurately on most trials, suggesting that the perceptual illusions of English speakers are partly due to their linguistic experience, rather than phonetic confusion alone. Further evidence against a purely phonetic explanation for our results is offered by the capacity of English speakers to perceive such onsets accurately under conditions that encourage precise phonetic encoding. The perceptual illusions of English speakers are also irreducible to several statistical properties of the English lexicon. The systematic misperception of universally dispreferred onsets might reflect their ill-formedness in the grammars of all speakers, irrespective of linguistic experience. Such universal grammatical preferences implicate constraints on language learning.     

The Sounds of Silence: Inferences from the Absence of Word‐of‐Mouth

While past work has explored some of the reasons why people themselves may remain silent in a group, almost no research has examined the mirror image of this question: How do consumers construe the silence of others? Do they project the opinions of the speakers in a conversation onto the silent individuals, assuming that silence signals agreement? Do they have a usual or “default” naïve theory of silence that they use to explain it across multiple contexts—i.e., “silence usually signals disagreement?” Or does silence act as a mirror, reflecting observers’ own opinions back at them? Three experiments contrasted perceivers’ estimates of conversational silence with their estimates of unknown opinions outside the conversation. Estimates of opinions outside the conversation generally followed an agreement‐with‐the‐speakers rule—the more an opinion was expressed in the group, the more consumers assumed others would support it too. In contrast, silence inside the conversation was interpreted very differently, serving as a mirror for participants’ own thoughts, even when the vocal majority favored the opposite position. Results suggest a process whereby observers project the reason they personally would have been silent in the group (given their opinion) onto silence, leading to an inference that the silents agree with the self.     

How children block learning from ignorant speakers

Preschool children typically do not learn words from ignorant or unreliable speakers. Here, we examined the mechanism by which these learning failures occur by modifying the comprehension test procedure that measures word learning. Following lexical training by a knowledgeable or ignorant speaker, 48 preschool-aged children were asked either a standard comprehension test question (i.e., “Which one is the blicket”) or a question about the labeling episode (i.e., “Which one did I say is the blicket”). Immediately after training, children chose the object labeled by an ignorant speaker when asked the episode question, but not when asked the semantic question. However, the advantage for episode questions disappeared when the same children were asked after a brief delay. These findings show that children encode their experiences with ignorant speakers, but do not form semantic representations on the basis of those experiences.     

Suprasegmental information affects processing of talking faces at birth

From birth, newborns show a preference for faces talking a native language compared to silent faces. The present study addresses two questions that remained unanswered by previous research: (a) Does the familiarity with the language play a role in this process and (b) Are all the linguistic and paralinguistic cues necessary in this case? Experiment 1 extended newborns’ preference for native speakers to non-native ones. Given that fetuses and newborns are sensitive to the prosodic characteristics of speech, Experiments 2 and 3 presented faces talking native and nonnative languages with the speech stream being low-pass filtered. Results showed that newborns preferred looking at a person who talked to them even when only the prosodic cues were provided for both languages. Nonetheless, a familiarity preference for the previously talking face is observed in the “normal speech” condition (i.e., Experiment 1) and a novelty preference in the “filtered speech” condition (Experiments 2 and 3). This asymmetry reveals that newborns process these two types of stimuli differently and that they may already be sensitive to a mismatch between the articulatory movements of the face and the corresponding speech sounds.     

Linguistic authority and convention in a speech act analysis of pornography

Recently, several philosophers have recast feminist arguments against pornography in terms of Speech Act Theory. In particular, they have considered the ways in which the illocutionary force of pornographic speech serves to set the conventions of sexual discourse while simultaneously silencing the speech of women, especially during unwanted sexual encounters. Yet, this raises serious questions as to how pornographers could (i) be authorities in the language game of sex, and (ii) set the conventions for sexual discourse—questions which these speech act-theoretic arguments against pornography have thus far failed to adequately answer. I fill in this gap of the argumentation by demonstrating that there are fairly weak standards for who counts as an authority or convention-setter in sexual discourse. With this analysis of the underpinnings of a speech act analysis of pornography in mind, I discuss a range of possible objections. I conclude that (i) the endorsement of censorship by a speech act analysis of pornography competes with its commitment to the conventionality of speech acts, and, more damningly, that (ii), recasting anti-pornography arguments in terms of linguistic conventions risks an unwitting defence of a rapist's lack of mens rea—an intolerable result; and yet resisting this conclusion requires that one back away from the original claim to women's voices being ‘silenced’.