Вы находитесь на странице: 1из 14

EFFECT OF COMMON GROUND ON LANG. PROD.

The Effect of Common Ground on Language Production Troy E. Gibson, Jr. Bloomsburg University of Pennsylvania

Author Note Troy E. Gibson, Department of Psychology, Bloomsburg University of Pennsylvania. Correspondence concerning this article should be addressed to Troy E. Gibson, Department of Psychology, Bloomsburg University of Pennsylvania, Bloomsburg, PA 17815. Email: teg16454@huskies.bloomu.edu

EFFECT OF COMMON GROUND ON LANG. PROD. Abstract

The purpose of this study was to determine if establishing common ground with a group member can affect the time it takes to complete a collaborative task involving descriptions. In this study there were 22 undergraduate students (20 female and 2 male) from Bloomsburg University of Pennsylvania. During the study the participants arranged themselves into pairs, 1 director and 1 matcher, and received 5 of the same tangrams of various shapes. For 3 trials the director arranged their tangrams in a random order and described the order to the matcher. Using the directors description the matcher attempted to accurately order their tangrams as fast as they could and record the completion time. Using the collected results we found that the average completion time for each trial decreased from Trial 1 to Trial 3. We concluded that this was because common ground was established between the director and the matcher which directly aided them in completing the task faster as the study progressed. Therefore, we concluded that establishing common ground can result in faster completion time in collaborative tasks. Keywords: common ground, language production, tangrams

EFFECT OF COMMON GROUND ON LANG. PROD. The Effect of Common Ground on Language Production Language production is a complex process. It involves various steps and incorporates social contexts considered by the speaker. Imagine a world where social context was not taken into account when trying to speak with other individuals. It can be assumed that language production without social context or rules as a filter would be harder to understand. This would occur not only because the listener would not comprehend the language, but also because the

speaker would not tailor their language for comprehension (Matlin, 2009). Therefore, social rules for language (i.e., common ground or directives) are important for the language production process. According to Matlin (2009), research by Clark & Van Der Wege (2002) shows that we start language production first with mentally planning out the gist of what we want to say. In other words we mentally decide the general message we are trying to articulate. Then research by Griffin & Ferreira (2006) shows we take the gist and form the basic structure to what we are trying to say and begin to figure out the exact wording we want to use by using the semantic situation as a filter (Matlin, 2009). Lastly, according to Matlin (2009) research by Treiman et al. (2003) shows that we take everything we mentally prepared and produce the appropriate phonemes to express it. At some point during this process speakers take into account social context to direct this language production process. As stated above, if this was not taken into consideration a speaker may not be understood by the listener (Matlin, 2009). Specifically speakers must think about who the listener is and what the listener knows so they are able to modify their speech for increased understanding (Matlin, 2009). In order for this to happen common ground must be established and utilized. Common ground refers to information shared between the speaker and the listener which provides an increased mutual understanding of any subject under discussion (Matlin, 2009). The focus of this study was on common ground and its

EFFECT OF COMMON GROUND ON LANG. PROD. effect on language production. Specifically we examined how establishing common ground

between a speaker and a listener can effect language production when describing abstract objects or tangrams. A similar study done by Horton and Keysar (1996) also examined how the knowledge of common ground between a speaker and a listener would affect the production of utterances. However, this study was unique because it explored the unexplained concepts about when common ground is applied to the language production process. The researchers looked at two models, Initial Design model as well as Monitoring and Adjustment model, in order to analyze if a speaker would apply common ground to the language production process before or after the utterance was made. For this particular study 24 University of Chicago undergraduates were recruited. Each participant was noted to speak English as their first language. The participants were randomly assigned to one condition from two sets of experimental conditions (i.e., privileged-context versus shared-context condition; speeded versus unspeeded condition) and were then asked to complete a modified referential communication task. During this task one participant and one confederate were placed on either side of a barrier dividing a computer screen. Then the participants were shown two varying objects (e.g., big and small or light and dark) and asked to describe the top object to the confederate before it moved to the other half of the screen (the confederate side of the barrier). Once the object appeared on the other half of the screen the confederate confirmed the participants description. The researchers recorded whether the participants descriptions utilized the second object which the confederate could also see or could not see depending on the participants condition. The researchers had found that the participants in the privileged-context unspeeded condition used common ground descriptions less than the participants in the shared-context unspeeded condition. However, the participants in the

EFFECT OF COMMON GROUND ON LANG. PROD. privileged-context speeded condition and the shared-context speeded condition used common ground descriptions equally. Overall, the researchers found that the participants did not rely on

the privileged-context information as much as the shared-context information. From these results the researchers concluded that people use the Monitoring and Adjustment model when utilizing common ground. They believed this would explain why there was no significant difference between reliance on shared or privileged information in the speeded conditions. Although within this model common ground is taken into account after the utterance is made, the researchers believed that in this study participants were able to reevaluate their utterances and eventually utilize only the ones with common ground. The researchers felt this explained why the participants in the unspeeded condition that had more time to reevaluate their utterances had a significant difference compared to the participants in the speeded condition that did not have enough time to reevaluate their utterances. The researchers also believed that individuals utilize the Monitoring and Adjustment model because using common ground requires too many mental resources to be done before an utterance is made. Therefore it would be more effective if the monitoring was done later in the language production process. One problem with this study was that the researchers used word descriptions to analyze common ground. It may be difficult to standardize the analysis of wording in order to determine if the participants were using common ground or not. The researchers did create a coding system, but it may have been flawed in determining if the participants had intended their description to use common ground. In the study done by Horton and Keysar (1996) they found that when two individuls communicate the speaker modifies their language during the language production process to match the common ground established between them. These findings however focus exclusively on common ground among two individuals. In contrast, a study done by Convertino et al. (2008)

EFFECT OF COMMON GROUND ON LANG. PROD. explored how common ground is produced and developed in a cooperative work environment

which involves more than two people. This study is different because it focused not only on how common ground is developed, but also on how common ground applies to teams which had never been done before. This study involved 36 participants total, two-thirds university students and one-third university employees with ages ranging from 20 to 45. The participants were randomly assigned to groups of three and were told they would have 20 min to solve a series of three different emergency evacuation problems. These problems involved providing a safety route out of an emergency situation that the researchers provided at random. Within the groups the individual members were given different expert roles (i.e., public works, environmental, or mass care) including shared and unshared information that must be used to complete the problem providing the best possible solution. After the group members were assigned individual roles they were given 10 min to thoroughly review the information provided followed by a 3 min assessment to ensure each participant adequately knew their information. During the study the groups were told not to reveal their unshared information to the other group members, but instead utilize that information combined with the shared information to collaboratively come up with a solution to the problem. The researchers found that the amount of time it took to complete the problems decreased from the 1st session to the 3rd session. They also found that the way in which the group members communicated altered by the 3rd session compared to the 1st session. The researchers believed this occurred because the participants began to employ more efficient means of conveying their information via common ground. These finding are significant because they provide an understanding of how groups communicate when doing complex tasks as well as the pros and cons associated with common ground in this type of communication medium. One flaw with this study is that it was not generalizable. The researchers tried to keep the study

EFFECT OF COMMON GROUND ON LANG. PROD. oriented towards real life situations of communication, but failed to recruit a wider variety of

participants. Conducting the study with only college educated participants analyzed the common ground effect among people that are more likely to have common ground previous to the study due to their education levels. In order to get a better understanding of the common ground effect, the researchers should have used more variability in the education levels of the participants. Overall, Horton and Keysar (1996) and Convertino et al. (2008) found similar patterns in the common ground effect. The researchers from both studies concluded that common ground is established between participants and that the participants did in fact alter their language in some way to fit the established common ground. Horton and Keysar (1996) specifically concluded that overall people rely on information that they know is shared between themselves and a listener when describing an object. Similarly Convertino et al. (2008) concluded that people change how they communicate in order to increase the comprehension among group members when solving problems. In this study we attempted to prove that common ground does affect language production. We chose to focus on the establishment of common ground when describing tangrams, or abstract objects. The participants were asked to complete a modified version of a study done by Clark and Wilkes-Gibbss (1986) as described by Matlin (2009). Our study involved the participants forming their own teams of two and designating a director and matcher which would describe the tangrams or arrange the tangrams respectively. The groups completed three timed trials in which the director described a random order of five tangrams to the matcher who attempted to accurately place their five tangrams in the described order as fast as possible. We predicted that by the 3rd trial the completion time would be faster compared to the 1st trial. This would occur because the director and matcher would establish a common ground on the description of the tangrams. Therefore, this would aid them in filtering out unnecessary

EFFECT OF COMMON GROUND ON LANG. PROD.

descriptions resulting in a faster completion time. By examining completion time we will be able to provide evidence that supports the common ground effect. Method Participants This study involved 22 undergraduate psychology students from Bloomsburg University of Pennsylvania. Among these participants there were 20 females and two males. These participants were asked to complete this study as part of the cognitive psychology course at the University. Therefore, they gave consent to participate upon review of the course syllabus. Materials The primary materials used in the study were a total of 110 tangrams. This included two sets of five varying groups with 11 identical patterns in each. One pattern from each group was provided to both the director and the matcher. During the study the director covered their set of tangrams with a large card which was provided and recorded the completion time with a pencil and paper which were not provided. The study was conducted in a standard class room in the McCormick Building on the Bloomsburg University campus, which included desks, a desktop computer, and an overhead projector. A standard stop watch was displayed on the overhead projector located in the room. Procedure The participants were asked to form their own groups of two. After all participant groups (11 total) were formed they were asked to designate one member as the director and one member as the matcher. Each participant group member received one tangram from all of the five tangram pattern groups. Overall, each participant group received one pair for each tangram pattern which was divided among them. The director was instructed to randomly order their

EFFECT OF COMMON GROUND ON LANG. PROD. tangrams in a different sequence for each trial without showing the matcher. Following this the director was instructed to describe the ordered tangrams one at a time to the matcher using only

descriptive words. Then, as quickly as possible, the matcher ordered their tangrams according to the descriptions provided by the director. Once the matcher accurately ordered their tangrams a completion time was recorded by the director for each trial. This study was repeated for three trials. Upon completion of the study all participants were debriefed. Results In this study we conducted a one-way repeated measures ANOVA on completion time during a test of common ground using tangrams. Analysis of the ANOVA revealed that there was a significant main effect of trial number on completion time, F(2, 30) = 21.169, p < .001. A Tukeys post-hoc test was performed on the effect of trial number on completion time. From this we concluded that there was a significant effect of trial number on completion time. This is because in Trial 1 (M = 25.64) participants had significantly slower completion times compared to Trial 2 (M = 10.70) and Trial 3 (M = 7.04). Also, completion times went down from Trial 2 to Trial 3. As shown in Figure 1, there is a downward trend in completion times from one trial to the next. Discussion In conclusion, we found that the average completion time in each trial decreased from Trial 1 to Trial 3. The downward trend in completion time suggests that common ground was established among the participant group members. These finding directly support our prediction that common ground would be established between participant group members resulting in faster completion times.

EFFECT OF COMMON GROUND ON LANG. PROD. Interpreting our findings we believed that common ground was established among the

10

participant group members, as stated above. In the 1st trial the group members had no common knowledge about how to describe the tangrams and therefore had slower completion times when ordering them. After completing the 1st trial the director and matcher began to recognize common descriptive words for each tangram. This assisted them in completing the task faster. In other words the director and the matcher established common ground. As a result of the established common ground between them, the director was able to provide better descriptive words to assist the matcher in ordering their tangrams at a faster rate. The interpretations about common ground on language production in this study were similar to the studies done by Horton and Keysar (1996) and Convertino et al. (2008). Although Horton and Keysar (1996) directed their study to locate when in the language production process common ground would if at all takes place, both of our studies provided the conclusion that common ground does modify language production. The main difference between this study and our study is that we provided completion time as evidence to support that common ground occurred between the participant groups. Instead Horton and Keysar (1996) put emphasis on the exact words used in order to provide evidence that common ground took place. Our findings contradicted the belief Horton and Keysar (1996) had about timed situations on the modification of language production. Despite their assumption that in a speeded situation common ground is less likely to take place, we both concluded that the common ground effect was the reason for our findings (Horton and Keysar, 1996). Just like Horton and Keysar (1996) we believed this was due to modification of descriptive words to increase understanding between the speaker and listener. Unlike the study done by Horton and Keysar (1996), the study done by Convertino et al. (2008) used completion time as evidence to show that common ground was established. This

EFFECT OF COMMON GROUND ON LANG. PROD. similarity to our study provided support that the completion times we collected could

11

demonstrate that common ground had an effect on the study. Despite this similarity, there were also differences between the study done by Convertino et al. (2008) and our study. The biggest difference is that Convertino et al. (2008) focused on groups instead of two individuals when trying to provide evidence to support the common ground effect. Also, instead of using abstract shapes such as the tangrams used in our study they utilized complex real life problems to evaluate the occurrence of common ground. Regardless of these differences, the study done by Convertino et al. (2008) as well as our study led to the conclusion that common ground caused modification of the language production between group members. Subsequently, this resulted in a faster completion time. There were several confounds for our study which may have altered our findings. First, the tangram shapes may not have been abstract enough. In other words they may have been easily identifiable by a simple description that would not require common ground. This may have led us to believe common ground was established even if it was not. Another confound could be the process of handing out the tangrams. During the study each tangram was distributed individually before the task began. This may have given some participant groups time to discuss what the tangrams looked like prior to the study beginning. Therefore we would not have been able to accurately determine if establishment of common ground during the study affected completion time. Also we did not randomize our participant groups; instead we allowed them to form groups on their own. This may have resulted in certain participant groups that worked well together and other participant groups that did not work well together due to previous relationships among them. This previous relationship may have affected how quickly common ground was established consequently changing our results. Lastly, our study utilized all

EFFECT OF COMMON GROUND ON LANG. PROD. psychology students. Because common ground is a concept discussed in the participants

12

cognitive psychology class, they may have had background knowledge about what our study was analyzing. As a result they may have been familiar with the common ground effect and modified their behaviors to best fit our predictions. In addition, one limitation to our study was that we used almost only female participants. In order to get a better understanding about how common ground affects language production among a general population we should have recruited more individuals of different genders. Overall common ground is an important concept to take into account for language production. Without thinking about who your listener is and your listeners knowledge it would be much more difficult to communicate effectively. Our study provided evidence to show that common ground does have a role in how we modify our language production. Specifically when trying to increase understanding between a speaker and listener during a collaborative task. To get a better understanding of common ground among varying populations future studies might include people that speak languages other than English as their first language or bilingually. Also future studies could examine how common ground differs between children and adults.

EFFECT OF COMMON GROUND ON LANG. PROD. References Convertino, G., Mentis, H. M., Rosson, M. B., Carroll, J. M., Slavkovic, A., & Ganoe, C. H.

13

(2008). Articulating common ground in cooperative work: Content and process. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (pp. 1637-1646). Florence, Italy: ACM. Horton, W. S., & Keysar, B. (1996). When do speakers take into account common ground? Cognition, 59, 91-117. Matlin, M. W. (Ed.). (2009). Language II: Language production and bilingualism. Cognition (7th ed) (pp. 324-334). Hoboken, NJ: Wiley.

EFFECT OF COMMON GROUND ON LANG. PROD.

14

30

25

Completion Time (s)

20

15

10

0 1 2 Trial Number 3

Figure 1. The effect of trial number on the completion time of a task utilizing tangrams to demonstrate common ground, including standard error bars. This shows that generally there is a downward trend in completion time for each trial. More specifically the completion time from Trial 1 to Trial 2 decreases significantly and the completion time from Trial 2 to Trial 3 decreases as well.

Вам также может понравиться