Conversation Cues & Selection of Information Sources Running Head: CONVERSATION CUES & SELECTION OF INFORMATION SOURCES Children’s and Adults’ Use of Conversation Cues When Selecting Sources of Information Sarah S. DeLisle Vanderbilt University 1 Conversation Cues & Selection of Information Sources 2 Abstract Word learning may be best characterized by the ability to recruit information from social others. One question, then, is how children decide to learn words from one person versus another. The present study investigates the possibility that children use a speaker’s ability to follow conversational norms when deciding whether to receive information about object labels. In particular, we investigated whether adults and four-year-olds are sensitive to violations of the Gricean maxims of quality and relation. Adults revealed sensitivity to violations of each maxim and used this to select from whom they received information. Four-year-olds were not sensitive to these violations. Future studies might pinpoint why children were unsuccessful in using the violations to decide whether to learn words from someone. Conversation Cues & Selection of Information Sources 3 Children’s and Adults’ use of Conversational Cues when Selecting Sources of Information Early word learning is quite amazing. In only a few short years a child’s vocabulary grows at a rapid pace. The skill with which children rapidly learn new words may be best explained in terms of their ability to recruit information from social others. Word learning is a social process; almost all of the information that children acquire comes from information that has been provided by another person (Bruner, 1990). Because children are learning information from social others, the quality of the information they learn is going to be determined by the quality of the source. Hence, one task children face is determining whether someone is a good source of information. Consequently, determining how children evaluate the quality of an information source is a question of importance. Discovering the answer to this question will provide insight into how children learn from social others. Past research has indicated that preschool children have some methods of determining whether someone is likely to be a good source of information. Research has indicated that children use the past accuracy of an informant when deciding whether to receive information from the individual. For example, when given the option of whether to receive new information from a past reliable or a past unreliable source, four-year-old children choose to receive information from the previously reliable informant (Clément, Koenig, & Harris, 2004). Additionally, when a previously reliable and previously unreliable information source both gave conflicting information, children chose to believe the information provided by the past reliable source (Koenig, 2007; Koenig and Harris, 2005). Furthermore, Koenig & Harris (2005) concluded that 3-year-olds are less systematic in their choice of likely reliable sources. In particular, 3-year-olds only Conversation Cues & Selection of Information Sources 4 showed a difference in learning from social partners when informants were consistently accurate; 3-year-olds would not learn from an informant who made a mislabeling mistake. Pasquini, Corriveau, Koenig, and Harris (2007) added to these findings by determining that 3-year-olds did not trust individuals who made a single error, while 4year-olds considered the number of errors in the context of the relative frequency of mistakes and, used this information to determine whether to trust the information that the individual provided. Thus, this research indicates that children around the age of four begin to reason more critically about the likely validity of an information source based on at least one aspect of that source. In particular, they use the past accuracy of the informant to determine whether they should learn from the person. In addition to children attending to the past reliability of an informant in determining the quality of the information that she provides, research also suggests that 4-year-olds are capable of attending to other related social cues. Koenig and Harris (2005) found that children consider cue words (e.g. “I know” versus “I think”) related to the blatant admission of knowledge or self-doubt in determining whether to learn the piece of information in question from the source. In particular, 3-and 4-year-olds consistently endorse labels offered by more confident speakers—speakers who use phrases indicating certainty (e.g. “I know”) as opposed to those offered by less confident speakers—speakers who used words indicating uncertainty (e.g. “I am not sure;” “I think”). Baldwin and Sabbagh (2001) also found that by four years of age children considered hesitancy of a source when determining whether to believe information provided. These findings, in conjunction with the findings on past reliability indicate that at around age four children are capable of judging the quality of an information source Conversation Cues & Selection of Information Sources 5 based on some aspects of verbal behavior in addition to the past reliability of an information source. Among the research on children’s evaluation of an information source, few studies have focused on other cues that characterize social interactions. Such studies for 4-year-olds are of particular interest because it is at this age that many children enter into a school setting where they have more social opportunity in number and in kind. As children begin to interact with a wider range of people in more diverse contexts, they may be forced to rely on a larger variety of cues when determining the likely quality of an information source. One set of cues that older children might come to use when determining the quality of information from a source is the ability of the source to follow conversational principles. There has been little research investigating the possibility that children attend to conversational principles when determining the likely correctness of the information that a source provides. Research indicates that individuals do create social representations of social partners (Spelke & Kinzler, 2007), but no research has yet tested if representations of conversational partners affect evaluation of the quality of information that a source provides. Such representations define a good social partner as an individual that follows social conventions such as joint attention, appropriate response to comments and questions, eye contact, and appropriate gestures. A bad social partner is an individual who does not adhere to such social conventions In the present study, the central research question is whether children and adults are more likely to learn words from “good” social partners or from “poor” partners. For this study, the creation of a “good” social partner and “poor” social partner in terms of Conversation Cues & Selection of Information Sources 6 conversational ability was developed from research by Paul Grice. Paul Grice (1975) developed a set of assumptions that script how conversational partners should interact with each other. Grice grouped these assumptions into the categories of quality, quantity, relation, and manner; he placed all of these categories under the general Cooperative Principle, which states: “Make your conversational contribution such as required, at the stage at which it occurs, by the accepted purpose and the direction of talk exchange in which you are engaged.” Each of the four maxims provides further insight into the meaning behind the Cooperative Principle. The maxim of quality states that speakers should say only what they know to be true. The maxim of manner addresses speakers to avoid ambiguity by being clear and succinct. The maxim of quantity says that speakers should say only as much as needed to be informative without saying more than what as necessary. The maxim of relation suggests that speakers should make each statement relevant to the context and to what has previously been stated. For this study, the ability of 4-year-olds and adults to evaluate the likely accuracy of information based on the source’s ability to adhere to the maxim of relation and to the maxim of quality was tested. These two maxims were selected in particular due to past research. Research by Eskritt, Whalen, and Lee (2008) found that children aged three to five become aware of these two maxims first. In this study children watched as two puppets gave hints to an experimenter about the location of a sticker which was hidden under one of four cups. One of the puppets violated a Gricean maxim while the other adhered to the maxim in the hint giving. Children were then asked to play the game and to choose one of the puppets to give them a hint about where the sticker was hidden Conversation Cues & Selection of Information Sources 7 under the cups. Children above chance chose the puppet that adhered to the Gricean principles of relation and quality. Since children at this age can recognize violations of relation and quality, we ask whether children at this age will also be able to apply such knowledge to make judgments about the information that a source provides in the context of word learning. We also conducted the study on adults, believing that if children and adults can identify good and bad social partners based on these Gricean principles, it might be likely that both age groups apply this information to determine from whom to trust new information. In the present study, we tested this in the context of children’s and adult’s word learning. Based on the previous research by Eskritt et al. (2008), we hypothesized that violations of Gricean maxims would be apparent to all participants. Additionally, using previous research about past accuracy and hesitancy (Clément et al., 2004; Koenig and Harris, 2005; Koenig, 2007; Baldwin and Sabbagh, 2001), we hypothesized that children and adults would be more likely to view individuals who adhere to the maxims as quality information sources and therefore, would more likely believe the novel label that they attributed to a novel object. Methods Participants Twenty-eight four-year-olds aged from 4;3 to 5;0 (M = 4 years; 6 months, 15 females, 13 males) and eighteen adults aged from 18 years to 22 years (19 females, 4 males) participated in the study. Child participants were typically developing, monolingual English speakers from predominately middle class families. Participants were recruited by phone from a university database that listed birth records of families Conversation Cues & Selection of Information Sources 8 interested in participating in research and residing in the Nashville, Tennessee area. Participants in the study received a small toy or a book as compensation. Adult participants were recruited through email and announcements in various classes at Vanderbilt University. Adult participants received a candy bar or course credit as compensation for their participation. Materials and Equipment In the study, participants sat on a love seat equidistant from two televisions. The participants were shown a different video on each of the televisions. One video featured a “bad” social partner violating the conversational maxim of relation or quality while interacting with a neutral partner. The other video featured a “good” social partner adhering to that same conversational maxim while interacting with the same neutral partner. The video featuring the violation of the maxim in the quality condition showed the bad social partner stating something that was not true. The video of a violation in the maxim of relation condition featured the bad social partner responding to the neutral partner’s questions with a random, unrelated comment. Each video interaction presented two clips of the actress violating or adhering to the maxim. Participants were shown each video twice. The two actresses who played either the good social partner or poor social partner were females named Megan and Teddi. Each interacted with the same neutral partner named Mark. Within each condition children either saw Megan as the good social partner and Teddi as the bad or vice versa. Teddi was featured in the context of playing with toy animals regardless of the condition or whether she played the good or bad role. Megan was featured in the context of playing with balloons regardless of the condition or Conversation Cues & Selection of Information Sources 9 whether she was featured as the good or bad social partner. See Table 1 for video scripts. Each of the videos lasted less than one minute. The participants (both 4-year-olds and adults) were introduced to four novel words given for eight novel objects. Therefore, every label was used on two objects. The novel objects included parts of toys that would not be known by the participant. The novel words included dake, glap, trome, teg. The novel words given for the objects were selected because they followed word patterns associated with English. See Figure 1. The novel objects were pulled from a shoe box decorated with construction paper and stickers that was referred to by the experimenter as a “toy box.” Design The study was designed to test the Gricean maxims of relation and of quality in two separate conditions. A between-subjects design was used. Four-year-olds were randomly assigned to one of the two conditions: Quality (N = 16; 9 females and 7 males) or Relation (N = 12; 6 females and 6 males). Adult participants were similarly randomly assigned to one of the two conditions: Quality (N = 11; 10 females and 1 male) or Relation (N = 12; 9 females and 3 males). For either condition participants viewed two videos. Participants in the Quality condition saw one video with a person violating the quality maxim and one video with a person adhering to the maxim. Participants in the relation condition saw one video with a person violating the quality relation maxim and one video with a person adhering to the maxim. Which actress (Megan or Teddi) was the good or bad social partner, the order the videos were shown, and the tv that each video was shown on were roughly counterbalanced across participants. Conversation Cues & Selection of Information Sources 10 After watching the videos, the participants participated in the word learning trial. The order for which the novel pairings/labels were presented was roughly counterbalanced so that each pair appeared in each position (1st, 2nd, 3rd, 4th) equally often. Procedure The session with child participants began with a 10 minute warm-up session. During this session the experimenter informed parents about the study, obtained consent, and played with a puzzle or drew a picture with the child. The warm-up session for adult participants involved the adult reading a description of the study and signing a consent form. After the warm-up session, the participant was asked to sit with the experimenter on a couch facing two televisions of equal distance from the participant. At this point, the experimental session began. The experimental session was divided into four phases—video introduction, video viewing, word learning, and memory check. Video Introduction. The researcher set the context for watching the two videos by telling the participant what was about to happen. When introducing the videos the experimenter established that the two actors were her friends—“Today we’re going to watch some of my friends on TV. Then we’re going to talk about the videos and play a game! On this TV we are going to watch Megan and Mark playing with balloons. Then on this TV we are going to watch Teddi and Mark playing with toy animals.” Participants were then instructed once more to pay close attention because a game would follow. Additionally, the experimenter stated the social partner’s name once again right before showing each of the videos (e.g. “Let’s watch Megan/Teddi and Mark”). Conversation Cues & Selection of Information Sources 11 Video Viewing. During the video viewing, children watched each video one after the other. After watching the first video once (which featured two clips of two separate examples of the social partner either adhering or disregarding the maxim) the experimenter said, “Let’s watch Teddi/Megan and Mark again.” The video was then shown a second time. Next, the experimenter said, “Now, let’s watch Megan/Teddi and Mark.” The process for the first video was then repeated. Word Learning. Immediately following the video session, the word learning phase began. The word learning phase featured four word learning test trials. The phase began with the experimenter asking the participant to identify which of the actors “knows more about stuff” and why. This question was designed to establish that participants had distinguished the good and bad social partners in the videos. Following this question, the word learning trials began. The experimenter set the context for the word learning test by stating that she and the two social partners in the videos had played with some toys the day before and that she was going to tell the participant what they (Megan and Teddi) had said about the toys in her toy box. After this introduction, four trials proceeded. Each trial consisted of the participant being introduced to two novel objects. When the two novel objects were presented, the researcher placed one of the novel objects in front of the TV in which the bad social partner had been presented (simultaneously, the TV now showed a full-screen picture of this social partner). The other object was placed in front of the TV where the good social partner had been featured (this TV now featured a full-screen picture of this social partner). While doing this, the experimenter gave the same novel name for each object and attributed the label for the different object to one of the social partners (e.g. Conversation Cues & Selection of Information Sources 12 “Teddi called this a glap” and “Megan called this a glap.”). After presenting the same label for two different objects and attributing the label to one of the social partners, the researcher closed her eyes and held out both hands, asking, “Can you give me the glap?” After the participant handed the researcher one of the objects, the researcher provided neutral feedback (e.g “ok”) and said, “Can you give me the other toy?” Both toys were then placed back into the toy box and a new trial began with the same procedure. Memory. The participant was then asked questions to ensure he/she had attended to the videos and remembered what happened during them. The participant was asked, “What was Megan playing with—balloons or dolls?” and “What was Teddi playing with—crayons or toy animals?” The participant was also asked, “Who would you rather play with? Why?” to investigate whether the participant had a preference for one of the social partners over the other. The entire session lasted around ten minutes. The sessions with the child participants were videotaped. Coding Our dependent measure was whether participants selected the novel object for which the good social partner had provided the label. Both child and adult participants received a score of 1 or a score of 0 for each of the four word learning trials. A score of 1 indicated that the participant had handed the object labeled by the good social partner to the experimenter after the request was provided. A score of 0 indicated that the participant handed the object labeled by the bad social partner to the researcher. There were a total of four trials; therefore, the maximum score value for the word learning trials was 4. Conversation Cues & Selection of Information Sources 13 Results To investigate whether children and adults are sensitive to violations of the conversational maxims of relation and quality a 2 (Age: 4 year olds vs. Adults) X 2 (Condition: Quality vs. Relation) between-subjects ANOVA was conducted on their tendency to select the object that the good social partner had labeled. The analysis revealed that there was no difference in the selection of the good social partner’s object based on the social convention that was broken (relation or quality) (F(1, 47) = .10, p = .76). Participants were equally likely to select the good social partner’s object in the quality (M = 2.94, SD = .22) and relation (M = 3.04, SD = .23) conditions. There was a difference in performance based on age (F(1,47) = 21.78, p < .001) with adults (M = 3.74, SD = .24) being significantly more likely to choose to the good social partner’s object than 4-year-olds (M = 2.250, SD = .22). There was no significant age X condition interaction (F(1,47) = .10, p =.76). See Figure 2. A second question concerned the reliability of a participant’s tendency to select the good social partner’s object. One-sample t-tests revealed adults’ choice of the good social partner’s object was above chance in both the quality (t(10) = 8.05, p < .001) and in the relation condition, t(11)= 11.00, p < .001. In contrast, one sample t-tests revealed 4-year-olds to be at chance in their selection of the good social partner’s object in the quality (t(15) = .67, p =.50) and relation condition (t(11) = .638, p = .54). These findings indicate that adults use a social partner’s ability to follow the Gricean principles of relation and quality to determine if the partner is a good source of information whereas children do not reliably use such conventions. Conversation Cues & Selection of Information Sources 14 A second look at the data was conducted to determine possible reasons for children’s poor performance selecting the labels provided by the good social partner. One initial possibility was that children simply did not remember the interactions (and the social partners) as well as the adults at the time of the word learning session. To ensure that memory was not the cause of poor performance, participants were asked questions in the memory phase about what each social partner was doing (e.g “What was Megan playing with—balloons or dolls?”; What was Teddi playing with—crayons of toy animals?”) All adult and 4-year-old participants responded to the memory related questions at the end of the word learning phase with 100% accuracy. Therefore, the difference in performance most likely cannot be explained by memory differences. If children were remembering the social interactions of the two partners, then there was another reason for the difference in child performance and adult performance. A second possibility for poor performance was that 4-year-olds could not identify which person “knew more” after viewing the videos. In order to investigate this, a chisquared test of association on the pattern of adults’ and of children’s responses to the question: “Who knows more about stuff?” was conducted. The chi-squared analysis indicated that there was a difference between the age groups in their responses, (χ² (1) = 21.64, p < .001). A look at the pattern of responding revealed that adults more often correctly identified the good social partner (22 out of 23 participants selected the correct person) than did 4-year-olds (8 out of 26 participants selected the correct person) as knowing more. Furthermore, the data indicated that children had a bias to attend to one social partner over the other when answering the “who knows more” question. A chi-squared Conversation Cues & Selection of Information Sources 15 test of association indicated that 4-year-olds were more likely to correctly identify the good social partner as the one who “knew more” when that partner was Teddi (χ² (1) 7.22, p = .007). Eighteen children who answered the question chose Teddi, with eight correctly choosing Teddi when she was the good social partner. Not a single child chose Megan regardless of whether she played the good social partner. Because this question was asked directly before the word learning phase, this may have meant that children were thinking about the wrong person (the person with whom they identified as “knowing more”) when they learned words. The 4-year-olds’ pattern of responses to the “who knows more” question provided an initial indicator that perhaps a social partner bias (Teddi preferred more than Megan) or context (toy animals preferred more than balloons) bias had emerged within the 4year-old test group. In order to further determine the presence of such a bias, a chi squared test of association for adult and child’s identification of who he/she “would rather play with” was conducted. The chi-squared analysis indicated that there was a relationship between age and preferred experimenter, χ² (1) = 5.25, p = .02. Children choose Teddi as a playmate 20 out of 23 times while adults chose Teddi 13 out of 23 times; this analysis supports the existence of a bias for 4-year-olds. General Discussion The current study investigated whether children and adults are sensitive to the Gricean maxims of quality and relation when determining the reliability of an individual as an information source in the context of word learning. The findings of this research suggest that adults are sensitive to violations of the conversational maxims of both quality and relation and rely on this sensitivity when determining from whom to receive Conversation Cues & Selection of Information Sources 16 new information. Adults not only identified an individual who followed these maxims as someone who “knows more” but also applied this knowledge when they were given conflicting information about the name of an object from an individual who followed the conversational maxim and an individual who disregarded the maxim. Furthermore, this study demonstrates that preschool age children are not reliably sensitive to violations of the conversational maxims of quality and relation. Four-year-olds neither reliably identified an individual who followed the conversational maxim of quality or relation as “knowing more” nor predominately believed that individual’s label for an object when provided with conflicting information from her and a social partner who violated the conversational maxim addressed. Four-year-olds poor performance in selecting the object in that the social partner who followed the conversation maxim of quality or relation (the “good” social partner) had provided the label was contrary to what we hypothesized. Our hypothesis was rooted in past research that suggests that children use aspects of an informant when selecting an information source. Research has indicated that preschoolers use an informant’s admission of ignorance, past mislabeling, and hesitancy to guide their beliefs about the reliability of the information source (Harris, 2007; Koenig et al., 2004; Pasquini et al., 2007; Sabbagh & Baldwin, 2001). Because children do use such aspects of an information source, it seems likely that they would use information related to conversational maxims. The fact that 4-year-olds did not at least use the maxim of quality when determining the reliability of an information source in our study is particularly striking because such a violation closely resembles mislabeling violations in which past research has suggested children are sensitive. Conversation Cues & Selection of Information Sources 17 One possibility for the surprising findings in our study in light of past research is that children in our study may not have even been recognizing the difference between the social partners in conversational ability so they could not apply this knowledge during the word learning trials. This is suggested by 4-year-olds’ performance on the “who knows more question.” Children were more likely to select Teddi as “knowing more” when she played the good social partner than Megan when she played the good social partner. Because past research has indicated that children at this age are most sensitive to the maxims directly addressed in this study— relation and quality (Eskritt, et al., 2008), the pattern of responses of the 4-year-olds suggests that there was an experimenter or context bias (the source of the bias cannot be determined because Megan was always featured in the balloon context and Teddi was always featured in the toy animal context). Since the “who knows more” question was asked directly before the word learning trials, it might be that this question focused the child’s attention on the wrong person, the person whom they preferred more. Adding further validation to the possibility of an experimenter bias is that 4-year-olds were more likely to choose Teddi than Megan when asked “Who would you rather play with?” A possible reason for the bias in answering the “who knows more” question is that children in our study were not able to detect and keep track of the violations of the maxims of relation and quality, and therefore, simply chose the social partner that they preferred more when answering. Future studies will seek to make violations clearer and will seek to provide more similar contexts featuring the example of the good and poor social partners; this will help to determine if making this violation more apparent will Conversation Cues & Selection of Information Sources 18 have an effect on children’s ability to differentiate ability between the two information sources. In sum, the present results suggest that 4-year-olds’ preference for one of the social partners in our study might have overridden their ability to differentiate between social partners who violate the conversational maxims of quality and relation and those who adhere to them. Future research is needed to investigate whether 4-year-olds use the conversational maxims of quality and relation at all to judge the information that a source provides. In order to answer this question, new videos should be designed that eliminate social partner or context bias. There is also a need to further investigate whether children simply are not able to complete the task presented in the study because of age. Perhaps, children may be able to recognize violations of the Gricean maxims of quality and relation in particular situations between the ages of three to five (Eskritt, et al., 2008), but that they may not apply this knowledge to choosing an information source for word learning until later. The current study shows that by the time an individual reaches adulthood, he/she considers maxim violations; this difference suggests that 4-year-olds may not be successful simply because of age. The difference in performance between 4-year-olds and adults suggests that sensitivity to violations of Gricean maxims is something that children eventually figure out. A future direction of the study would be to determine the age at which children develop this ability to differentiate reliable information sources based strictly on violations of conversational cues. Discovering the age at which children develop the ability to both establish a source as a “good” or “bad” conversational partner and then to apply this knowledge to Conversation Cues & Selection of Information Sources 19 evaluate the quality of information from that source has many practical implications for children’s learning. Recent research in the education field supports the theory that children consider cues related to the ability of conversational partners when learning in the school context. Research out of the educational field by Pianta, Belsky, Houts, and Morrison (2007) suggests that children’s emotional engagement predicts their willingness to learn in the classroom. Specifically, this research indicates that young children’s learning is affected by the quality of their interactions. Therefore, if children evaluate social partners in the ways indirectly suggested by the findings in Pianta, Belsky, Houts, and Morrison’s research, determining the age at which children begin to evaluate the quality of interactions based on an information source’s ability to follow social conventions would have many implications for how teachers conduct themselves in the classroom. Conversation Cues & Selection of Information Sources 20 References Bruner, J.S. (1990). Acts of meaning. Cambridge, U.K: Cambridge University Press. Clément, F., Koenig, M., & Harris, P. (2004). The ontogenesis of trust. Mind &Language. 19; 360-379. Eskritt, M., Whalen, J.& Lee, K. (2008). Preschoolers can recognize violations of Gricean maxims. Britist Journal of Developmental Psychology, 26, 435-443. Grice, H.P., (1975). Logic and conversation. In P. Cole & J.L. Morgan (Eds.), Speech acts, syntax and semantics3 (pp. 113-128). New York: Academic Press. Harris, P.L. (2007). Trust. [Electronic Version]. 10; 135-138. Koenig, M.A. & Harris, P.L. (2005). Preschoolers mistrust ignorant and inaccurate speakers. Child Development. 76; 1261-1277. Koening, M.A., Clement, F., & Harris, P.L. (2004). Trust in testimony: children’s use of t true false statements. Psychological Sciences. 15; 694- 698. Pasquini, E.S., Corriveau, K.H., Koenig, M., Harris, P. (2007). Preschoolers monitor the relative accuracy of informants. Developmental Psychology. 43; 1216-1226. Pianta, R.C., Belsky, J., Houts, R., & Morrison, F. (2007). Opportunities to learn in America’s elementary classroom. Science Magazine. 315; 1795-1796. Spelke, E.S. & Kinzler, K.D. (2007) Core Knowledge. Developmental Science, 10; 8996. Sabbagh, M.A. & Baldwin, D.A. (2001). Learning words from knowledgeable versus ignorant speakers: links between preschoolers’ theory of mind and semantic development. Child Development. 72; 1054-1070. Conversation Cues & Selection of Information Sources Figure Caption Table 1. The video scripts for each condition and each social partner. Figure 1. The novel objects pairs with the accompanying label for each. Figure 2. A graph of the Mean Performance for Age and Condition. 21 Conversation Cues & Selection of Information Sources 22 Quality Condition Scripts: Neutral Do you want to play a game with me? “How many balloons do you have there? “What color is that one?” (pointing to a blue balloon) Neutral Do you want to play a game with me? “What do you want to do with the animals?” “What animal is that one?” (pointing to a pig) Meg Good Meg Bad Sure Sure “I have 2 balloons here.” “I have 60 balloons here.” “That one is blue.” “That one is white.” Teddi Good Teddi Bad Sure Sure “Let’s put them in this box.” (pointing to a box) “That is a pig.” “Let’s put them in this cup.” (pointing to a box) “That is a duck.” Relation Condition Scripts: Neutral Do you want to play a game with me? “How do you want to play with the balloons?’ “I don’t like to play with red balloons” Neutral Do you want to play a game with me? “What do you know about these animals?” “I don’t like to play with toy snakes?” Meg Good Meg Bad Sure Sure “Well, I can throw them to you and you can throw them back.” How about green balloons? Do you like to play with green balloons? “Yes, I can use a fork and a spoon to eat my spaghetti.” Teddi Good Teddi Bad “How about red balloons? Do you like to play with red balloons?” Sure Sure “Well, I know that a pig says oink and a duck says quack.” “What about toy dogs? Do you like to play with toy dogs?” “Yes, I can use a bat and a ball to play baseball.” “What about toy snakes? Do you like to play with toy snakes?” Conversation Cues & Selection of Information Sources Glap vs. Glap Teg vs. Teg Dake vs. Dake Trome vs. Trome 23 Conversation Cues & Selection of Information Sources Mean # Correct Mean Perfomance For Age and Condition 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0 3.833 3.636 quality relation 2.25 2.25 quality relaiton Adults 4 Year Olds Age X Condition 24