The Daily Use of Imagined Interaction Features

Imagined interactions (IIs) refer to a process of social cognition in which individuals imagine, and therefore, indirectly experience themselves in anticipated and/or past communicative encounters with others. In this manuscript, two groups of participants kept diaries of their daily IIs. In the first study, rehearsal and proactivity were the most frequently reported features. Half of the entries reported only one function for the II. In the second study, catharsis and proactivity were the most frequently reported features, and there were positive associations between the number of II functions and II attributes featured in the diary and the interpersonal cognitive complexity of the participant. The contributions of these findings are discussed in regard to II theory and intrapersonal communication.

2003). Commonly, II research has focused on understanding how IIs relate to other concepts, yet research should also strive to understand II features as they relate to each other. Whereas some recent quantitative work has investigated II features and their interrelationships (e.g., Bodie, Honeycutt, & Vickery, 2013), there is room to investigate these features as they are reported in other forms. For instance, IIs can be shared and recounted through oral histories, diary entries, interviews, and other narrative accounts.
In this manuscript, the features of IIs reported in diary entries were addressed in two studies designed to investigate which features are most frequently reported in daily accounts of IIs. Furthemore, in an effort to understand better the multidimensional nature of IIs, study 1 examines how individuals report using single or multiple functions in their daily IIs. Study 2 examines the relationship between interpersonal cognitive complexity in terms of construct differentiation and the use of multiple features across diary entries. Both studies advance understanding of how individuals think about interactions with others, and provide support for the ongoing, continuous nature of our imagined interactions. Following is a brief review of the features of imagined interactions including the use of different methods for studying them.

Imagined Interactions and Intrapersonal Cognition
Imagined interaction theory is based in the work of symbolic interactionists, including Mead (1934), who discussed the internalized conversation of gestures in which individual actors monitor social action by reviewing the alternative endings of any given act in which they are involved. Individuals use internal dialogues to mentally test out the various possible scenarios of an event in advance of the act. IIs have their foundations in cognitive scripts (Honeycutt & Bryan, 2011). Relational scripts are partly formed through the process of mental imagery and daydreaming, in which individuals often think about conversations with significant others. II theory rests on the assumption that intrapersonal communication, or internal talk, is the "foundation on which other types of communication rest" (Honeycutt, 2008, p. 79). The importance garnered by internal talk has sparked curiosity regarding its fundamental features, which are currently described by two distinct categories: II functions and II attributes.
In generating intrapersonal cognitive representations of conversations, IIs are seen as "possessing many of the same characteristics as real conversations in that they may be fragmentary, extended, rambling, repetitive, or coherent" (Honeycutt, 2003, p. 13). Indeed, in the study of conversation, any singular conversation can be described and analyzed from a variety of perspectives, including the primary or secondary goals enacted in the conversation (Dillard, Segrin, & Harden, 1989) and the various topics or types of talk present in the conversation (Goldsmith & Baxter, 1996).
The six functions and eight attributes articulated here represent how IIs are investigated from the perspective of II theory, and do not illustrate every possible way to investigate intrapersonal social cognitive processes. There are six functions of IIs that explain motivations for their usage: (1) catharsis, or using IIs to relieve tension and reduce uncertainty; (2) compensation, or substituting IIs for real interactions; (3) conflict-linkage, or using IIs between arguments, thereby creating an ongoing conflict; (4) rehearsal, or mentally preparing for interactions prior to engaging in them; (5) relational maintenance, or using IIs psychologically to maintain a relationship by thinking about a partner; and (6) self-understanding, or using IIs to gain a deeper understanding of the individual's attitudes, values, and beliefs (Honeycutt, 2010). Recent research has examined physiological arousal in terms of rumination and conflict-linkage, as well as recall of positive family stories (Honeycutt, 2010;Honeycutt, Bannon, & Hatcher, 2014;Honeycutt, Keaton, Hatcher, & Hample, 2014).
There are eight attributes or characteristics of IIs that have been associated with numerous dispositional variables, including narcissism, communication apprehension, and the Big 5 personality traits (Choi et al., in press;Honeycutt et al., 2009;Honeycutt, Pence, & Gearhart, 2013). Early II research examined sex differences and associations with personality attributes, marriage types, and relational quality (for a review, see Honeycutt, 2003).
Frequency simply represents individual differences in how often individuals experience IIs. Proactivity and retroactivity are concerned with the timing of the II in relation to actual conversations. Proactive IIs occur before an anticipated encounter, whereas retroactive IIs occur after the encounter. Retroactivity is very common in films and movies in which characters are shown having flashbacks. Proactive and retroactive IIs can occur simultaneously as individuals replay prior conversations in their minds while preparing for ensuing interactions. Discrepancy occurs when what was imagined is different from what actually occurs in conversations. The self-dominance attribute reflects the tendency for most of the imagined talk to originate from the self with less emphasis being placed on what the interaction partner says. The variety characteristic of IIs reflects individual differences in the number of topics that are discussed in the IIs and whom they are with. IIs tend to occur with significant others such as relational partners, family, and friends, rather than with strangers or infrequent acquaintances. Valence, the seventh attribute of IIs, refers to the amount and intensity of emotions that are experienced while envisioning conversation (Honeycutt, 2010). Finally, IIs vary in their specificity, or how vague the imagined lines of dialogue are, as well as the setting where the imaginary encounter occurs. Some people are specific in using verbal imagery as they recall lines of dialogue, whereas others can detail the exact scene of the interaction in which visual imagery is accented.

Testing II Theory: Common Methodologies
Most research on IIs has been conducted using self-report survey methods, including both the short and long form Survey of Imagined Interactions (Honeycutt, 2003(Honeycutt, , 2010. Because IIs are a cognitive phenomenon, this methodology allows researchers to capture the uses and descriptions of IIs as reported by individuals who use them. When using this self-report methodology to examine IIs, researchers have made modifications to the Survey of Imagined Interactions (SII) guided by theoretical and practical concerns. For instance, the SII has been modified to compare the use of specific functions across varying cultures (McCann & Honeycutt, 2006), with the remaining functions excluded from the survey and subsequent analysis. In another study exploring conflict, only the function of conflict-linkage was included as a way to understand the correlates between conflict-linkage IIs, rumination, and taking conflict personally (Wallenfelsz & Hample, 2010). The SII has also been adapted to assess the corresponding relationships between II functions and attributes (Bodie et al., 2013). The SII is appropriate for assessing the propensity to engage in the individual features of IIs (functions and attributes) and does have an open-ended section focused on a recent, recalled II. Nonetheless, the SII instrument is limited in that it does not account for the frequent and habitual use of II features. Using a method that allows for this possibility may provide additional insight into past research focused on the features of IIs, as well as provide opportunities for future, qualitative research into IIs based on these findings. Despite a general lack of research focused solely on refining and expanding II theory in and of itself, the SII and modifications made to this instrument tend to explore IIs in relation to other concepts, rather than expanding knowledge of IIs as a distinct theory. Thus, one major goal of this study is to assess how frequently the features of IIs occur when participants are not constrained by the SII's focus on the general occurrence of features and are instead able to record their II usage more freely through the use of diaries.
The present work's chosen method is not without precedent. Diary accounts (Edwards, Honeycutt, & Zagacki, 1989) and oral histories (Honeycutt, 1999) have been utilized in II research, but not as commonly as the SII. Denzin (2009) discusses how each method reveals different aspects of empirical reality and recommends triangulation through incorporating multiple methods of analysis. Therefore, we used log diaries in the present studies. Whereas diaries and logs are both self-reports, they represent the immediate recording of experiences, unimpaired by reconstructions and distortions of memory (Denzin, 2009). As part of diaries, the log is a list of imagined interactions and provides narrative accounts of what an II is typically like in daily life. By focusing on daily experiences of IIs, this study contributes to testing II theory and may bring forth additional insight into the nature of IIs currently unexplored by methodologies relying on survey instruments.
The recent quantitative research by Bodie et al. (2013) has advanced II theory by using the SII to produce some unexpected results. For instance, these researchers found that, contrary to their predictions, relational maintenance and conflict IIs were used just as frequently as those for catharsis. Their findings introduce new areas for testing II theory-most importantly, exploring, attributes and functions of which are used most frequently over an extended period of time. Based on these advancements, and on the need for tests of II theory to employ methodologies that capture best the particularly salient questions, this manuscript focuses on the long-term use of II features and attributes. This focus is reflected in three research questions: RQ1: What function is most reported over extended lengths of time? RQ2: Which attribute is most reported over extended lengths of time? RQ3: Do multiple (or singular) functions of IIs occur in diary accounts?
This manuscript presents two studies, conducted in consecutive semesters with different participants, aimed at understanding better the frequency and use of II features. The studies follow similar methods to investigate the central research questions.

Method
Participants. Participants were students (N = 36) enrolled in an upper-level communication studies course at a large university in the southeastern United States. The average age of participants was 20.34 years (SD = 2.56). A slight majority of the sample was female (53%). Most participants reported being from a middle-class socioeconomic background and a majority of the sample (n = 27, 75%) reported identifying as Caucasian, but Black (n = 5, 13.88%) and Hispanic ethnicities (n = 4, 11.11%) were also represented. Each person submitted one diary log that contained all entries. Participants were informed that their diaries would be used for research purposes following IRB procedures, but their name and any identifying information would not be associated with their entries. Participants received a portion of their class credit for completing the assignment.
Procedure. Participants were instructed on the first day of class to start a diary recording their imagined interactions and intrapersonal communication. They were instructed to ask any follow-up questions about the assignment during class, as well as, via email with the instructor. From the initial assignment date to the assignment deadline, the students kept their diaries for 79 days (11 weeks and 2 days) during the 16-week semester. The following instructions were provided in the course syllabus to guide their entries: Write about your intrapersonal communication and imagined interactions. Where do your imagined interactions occur? How often? Who are they with? Where are you when you have them (e.g., in your residence, driving a car, walking across campus, using the Internet)? What are the imaginary topics of conversation? What is said? Additionally, write them out as soon as possible and describe other daydreams or fantasies, feelings about yourself, your goofs in speaking and understanding your experience of stimuli, self-generated or externally produced.
At the end of the 79-day period, participants turned in a single packet or notebook (referred to as the "diary" unit in this manuscript) containing their handwritten or typed entries of intrapersonal communication recorded during this time period.
The diaries were randomly numbered for analysis. Examples and excerpts from the diaries included in this manuscript omit names of II partners and specific locations.

II Feature Identification and Coding Procedures
General procedures. The entries were coded for II features. For the coding scheme, definitions were discussed in an initial training between all the authors, and following this training the second author provided a finalized coding scheme document to the third author. 1 There were two primary components in coding II entries. The first component was identifying whether an entry qualified as an Imagined Interaction. Entries were determined to contain IIs if the intrapersonal communication being described involved another individual or group of individuals, but were otherwise coded as self-talk if there was no interaction partner. After IIs were identified, the second component consisted of coding the II features (attributes and functions) present in the entry. For some entries, participants explicitly identified features using the theoretical terms of II theory. For instance, II attributes such as specificity were clearly named (e.g., "This II was very specific"). In other entries, the features were clearly described but without the theoretical term present-for instance, discrepancy (e.g., "This II was not at all like what happened in the real conversation") and proactivity (e.g., "I had this II before my interview"). Reliability of coding IIs was assessed for (1) agreement on whether the entry contains an II or not; (2) agreement on the list of II features (functions and attributes) named by the participant in the entry; and (3) agreement on the list of II features (functions and attributes) described but not explicitly named by the participant in the entry.
The second and third authors selected four diaries at random (11.1% of the total sample) to serve as a representative sample for coding. 2 The authors coded the four diaries independently using the coding scheme. First, for agreement on the presence of an II in the entry, there was 100% agreement that all entries coded (n = 79; 10.6% of total II entries) contained IIs and each entry contained only one II. 3 Second, for listing the II features identified by the participants, there was 100% agreement and accuracy in listing these participant named II functions and attributes (n = 151). Finally, to account for those II features described but not explicitly identified, Krippendorff's alpha coefficient was utilized to determine the degree of agreement between authors. Reliabilities were appropriate for II attributes (α = .77) and II functions (α = .73). To ensure the accuracy of coding the features not explicitly identified by participants, the authors met and discussed any disagreements before coding the remaining diary entries (n = 890 in 33 diaries). These entries are the primary unit of analysis in analyses addressing the research questions.
Generating frequency counts. After the entries were coded, all II features were tabulated to answer the research questions. The frequency of all features was tabulated for each entry. For the functions, all six functions were tabulated as described in the literature review and in the coding scheme and are represented by frequency data. The attributes of proactivity and retroactivity were also tabulated as described in the literature review and the coding scheme. However, due to the nature of the attributes of valence, dominance, specificity, and discrepancy, the tabulations of these four features are subdivided into eight total categories, representing the conceptual continua that undergird these attributes. For valence, entries were tabulated into two separate categories representing positive and negative valence; for self-dominance, entries were tabulated into two categories representing selfdominance and other-dominance; for specificity, entries tabulated into two categories representing low specificity (vagueness) and high specificity; and for discrepancy, the entries were tabulated into either low discrepancy (very similar to the interaction) or high discrepancy (very different from the interaction) categories.
Coding for frequency and variety. The attributes of frequency and variety were tabulated differently due to the nature of these features as general descriptions of the use of IIs and the content within these IIs. To tabulate frequency, the total number of II entries was tabulated and summed for each diary, and this number was then divided by the total number of days the participants kept the journals, in order to represent the frequency of IIs. To tabulate variety, the total number of unique, nonredundant II partners specified and the total number of unique, non-redundant II settings specified were tabulated and summed for each diary.

Results
The first study was undertaken to investigate the central research questions related to the use of II features over time. The diary entries were the unit of analysis as these entries reflect the features present in IIs. There were totally 969 entries, and 76.5% of the entries (n = 742) qualified as II entries and the remaining analyses focus on these 742 entries. The excluded entries (23.4%, n = 227) contained self-talk, daydreams, fantasies and other aspects of intrapersonal communication in accordance with the original assignment instructions.
Diaries contained on average 26.64 total entries (SD = 9.27; range 12-51 total entries); the average number of II entries per journal was 20.61 (SD = 9.63). These entries report IIs employing the six identified IIs functions. We report the results in the order of the three central research questions. RQ1: What function is most frequently reported over extended lengths of time? There were 1,092 identified uses of functions in the II entries. Rehearsal was the most frequently reported function (n = 349, 47.0%) and the least frequently reported function was compensation (n = 68, 9.2%).
RQ2: What attribute is most frequently reported over extended lengths of time? There were totally 1,213 identified uses of attributes in the entries. Proactivity was the most frequently reported attribute with over 58% of the entries reflecting this attribute, and discrepancy was the lowest reported attribute with almost 4% reporting this attribute.
As variety and frequency were tabulated differently than the other attributes, it is important to report the details of these results in relation to this second research question. Furthermore, variety was recorded to reflect both the variety of imagined conversational partners, as well as imagined II settings. We report all measures of central tendency. The number of unique partners featured in II entries in a single diary ranged from 6 to 29 (M = 15.19 partners; SD = 5.66; Mdn = 15 partners, Mode = 16 partners). The number of unique imagined settings featured in II entries in a single diary ranged from 0 specified settings to 21 specified settings (M = 8.22 settings, SD = 5.43; Mdn = 9 settings; Mode = 9 settings). Finally, frequency was represented by the average number of II entries each day. This number was generated by taking the total number of II entries for each diary and dividing the total entries by the number of days (79 days) the participants kept the diary. On average, participants recorded fewer than one II per day (M = .26 entries per day, SD = .12, Mdn = .23, Mode = .23).

RQ3: Do multiple or singular functions of IIs occur in diary accounts?
The third research question of the first study focused on the multifunctional nature of IIs. To answer this research question, the II functions featured in each entry were tabulated to determine which entries contained singular or multiple functions. We found that 56.20% (n = 417) of IIs reported only one function. With these singular II function entries, rehearsal had the highest frequency count for singular functions with 167 (40%) of the single function entries. Although entries mentioned up to four functions, entries featuring one function were most frequent, followed by dualfunction entries (37.33%; n = 277). In comparing the frequency distribution between IIs with one, two, three, or four functions, the difference between categories was statistically significant, χ 2 (3) = 648.95, p < .01. In response to research question three, single-function II entries were the most common.

Discussion
With 742 entries devoted to IIs, these entries revealed rich information about the experiences of daily IIs. In this study, the most frequently reported function was rehearsal and the most frequently reported attribute was proactivity, answering the first and second research questions. As the second study shares these research questions, a more thorough analysis of the first two research questions is discussed in the combined discussion section.
The third research question focused on discovering whether participants report using multiple functions in a single II. As Honeycutt (2003Honeycutt ( , 2008 notes, II functions may occur simultaneously. For example, the person may experience catharsis while thinking about how to manage a conflict for an upcoming conversation. In exploring this research question and general theoretical claim about II theory, we found that over 50% of IIs reported only one primary function. It is possible that when recalling an II, only the most salient features are recalled. That is, an II may have been used for catharsis, self-understanding, and rehearsal, but only one or two of these features are recalled and highlighted when the II is written down in the diary. Indeed, entries contained high specificity only 57 times in the entire sample (9.01% of all attributes specified), supporting the idea that, in general, the entries reflect a broader report of IIs versus a highly specific and detailed account of IIs. People often remember more emotional events than non-emotional ones across a variety of visual and verbal stimuli, as well as autobiographical recall (Kensinger & Schacter, 2008). Since catharsis is associated with emotional release, recall of catharsis and understanding is congruent with the emotional recall studies in cognitive psychology. Moreover, the valence of recall is comparable for positive and negative events.
This study adds to our understanding of how frequently features are reported in IIs: Individuals are likely to engage in focused IIs that are more useful and require less effort on behalf of the individual. These focused IIs allow for better understanding of events that have occurred between friends or family (e.g., selfunderstanding and relational maintenance), as well as, allow for better preparation for upcoming events through exploring emotions and possible scripts for an interaction (e.g., rehearsal and catharsis). The most common way to describe IIs is relative to the timing of conversations, supporting the use of IIs as a way to complement actual interactions. IIs are used throughout our daily lives, with imagined conversations preparing individuals for upcoming actual interactions.

Study 2: The Use of II Features and Interpersonal Construct Differentiation
There is a variety of potential explanations for the frequencies that were observed in Study 1 in support of the usage of multiple II functions. For instance, whereas the data do not reflect how specific individuals use IIs, we noted that many entries reflected a propensity to use IIs for a particular function. This pattern presents an interesting theoretical question: Why do individuals vary in their patterned use of specific II functions? Differences in interpersonal cognitive complexity are one explanation. Individuals vary in their ability to represent and process social information and this difference may influence how frequently an individual engages in IIs and what various features are activated in these cognitions. Research has revealed contrasts in individual personality differences and IIs (e.g., Honeycutt & Keaton, 2012), as well as, differences between the use of IIs, conversational sensitivity, and communication competence (Honeycutt et al., 1992(Honeycutt et al., -1993 but there may be other explanations for the propensity to use multiple II functions and to report using a variety of II features in imagined conversations. One difference may be in the ability of individuals to identify and organize social knowledge in more or less efficient and useful ways. Based on the findings in Study 1 reflecting the varied use of II functions, particularly the use of IIs primarily for one function, Study 2 seeks an explanation based in individual differences in functional communicative behavior for the varied use of II features. The constructivist perspective represents "a general approach to communication with applicability to a wide range of specific phenomena" (Delia, O'Keefe, & O'Keefe, 1982, p. 165), with specific advantages for examining differences in functional communicative behavior including message production, message processing, and interaction coordination. According to this perspective, individuals vary in their underlying cognitive structures regarding communication, called interpersonal constructs. These underlying structures, or constructs, are described by Delia, Burleson, and Kline (1979) as "the schemes within which others' behaviors are interpreted, evaluated, and anticipated" (p. 244). This definition of constructs provides some initial parallels between the II functions of self-understanding and rehearsal. An individual's interpersonal construct system can vary in terms of the number of constructs, the abstractness of these constructs, and the integration of these constructs.
Interpersonal cognitive complexity represents the number of psychological constructs that people use to describe others, and it has been linked across a variety of communication domains, including crafting supportive messages, understanding others, empathizing, and making situational attributions (Bodie et al., 2011;Burleson & Caplan, 1998). Individuals high in cognitive complexity have increased information comprehension (Beatty & Payne, 1985) and increased conversational recollection (Neuliep & Hazleton, 1986), so they may have IIs with greater specificity of verbal imagery and may also have more accurate recollection of prior conversations in retroactive IIs. Rehearsal IIs have been to reduce the number of object adaptors in subsequent conversations (Allen & Honeycutt, 1997), suggesting that IIs can function in adapting communicative behavior in conversation, and suggesting a need to investigate other individual differences affecting conversational activity and involvement.
Focusing further on cognitive complexity as an indicator of the number of social cognitive constructs available and accessible when planning for and engaging in conversation, we explore the functional communicative behavior of interpersonal construct differentiation within the system of social constructs. An individual's ability to differentiate constructs is shaped through social experiences (e.g., Delia et al., 1979), reflecting social cognitive development through the identification and organization of information gleaned from social experiences with others. Interpersonal construct differentiation is an important area of focus within this larger domain of research, as individuals high in construct differentiation may rely on differentiated constructs in the generation of IIs rehearsing, analyzing, and reliving conversations. Based upon message production studies, we would tend to predict that IIs would be proactive (Waldron & Applegate, 1994) and would be used for rehearsal based upon the individual's greater metacognitive awareness. O'Keefe and Delia (1988) as well as Kline and Floyd's (1990) research on message production reveal how skills at message design and adaptation (e.g., refusing requests) improve with maturity. Relatedly, persuaders with highly developed constructs design messages that emphasize the connection between beliefs, goals, and feelings compared to less differentiated persuaders (Kline, 1991).
Message design is reflected in numerous functions of IIs, including rehearsal, conflict-linkage, and catharsis in terms of designing messages for emotional release.
For instance, as individuals have more differentiation in their imagined conversations, they may be more likely to engage in IIs that more accurately reflect the multifaceted nature of conversations, invoking high levels of detail (specificity) and imagining what another person might say (other dominance), while relieving negative emotions (catharsis) and preparing for an upcoming conversation (rehearsal). This pattern of engaging in more functions and attributes may then be observed over a longer period of time, with individuals reporting a wide variety in the functions of their IIs. Other individuals may not enact a variety of features, instead relying on IIs to rehearse for a conversation, or always having IIs prior to seeing a person. These individuals may then only repeatedly use one or two II functions versus achieving a more developed and differentiated approach to internalized conversations. This difference prompts a final research question for Study 2: RQ4: Is there an association between interpersonal construct differentiation and the use of II features over an extended period of time?
Additionally, to capture the association between interpersonal construct differentiation and the use of II features, the authors coded participants' reported identification of II entries as reflecting frequency and variety, differing from how these attributes were coded and tabulated in the first study. In this second study, frequency and variety are treated as equivalent to the other attributes of IIs in terms of coding.

Method
Participants. Similar to Study 1, participants were students (N = 37) enrolled in an upper-level communication studies course at a large university in the southeastern United States. The mean age was 20.13 years old (SD = .48). The majority of participants were female (65%). The ethnic composition and socioeconomic status were similar to that of the Study 1 participants. Participants were informed that their diaries would be used for research purposes, but their name and any identifying information would not be associated with their entries. Participants received a portion of their class credit for completing the assignment. Study 2 occurred in a different semester than Study 1 and no participant was involved in both studies.
Procedure. The procedure was very similar to Study 1, but instructions were modified to include two clauses encouraging participants to report on if their IIs reflected frequency and variety. Participants were instructed on the first day of class to start a diary recording their imagined interactions and intrapersonal communication. From the initial assignment date to the assignment deadline, the students kept these diaries for 82 days (11 weeks and 5 days) during the 16-week semester.
When turning in diaries and when completing the Role Category Questionnaire (RCQ) to assess interpersonal construct differentiation, participants used their student ID number. Only the first author was given access to this information, and once the diaries and forms were matched, a new participant number was randomly generated and the student ID numbers blacked out to ensure confidentiality.

II Feature Identification and Coding Procedures
Coding procedures similar to those in Study 1 were used in the second study to code the II features in the diary entries. The authors met to train on coding procedures and reviewed example diary entries from the first study. One modification in Study 2 was in capturing the II attributes of frequency and variety. In Study 2, no additional coding of partners, settings, or number of entries were tabulated. These attributes were coded using only explicit references to frequency and variety in the entries with no additional coding of these features.
The authors then independently coded three diaries (8.1% of the total diaries), following the coding procedures outlined in the first study. The coders had 100% agreement on the number of entries that included IIs (n = 62; all entries in the sample) present in three randomly selected diaries, as well as 100% agreement when listing participant identified features. Krippendorff's alpha was also excellent, reflecting high interrater reliability for both II attributes (α = .96) and II functions (α = .98). The authors coded all remaining entries in the 34 remaining diaries.
For all diaries, the authors read through all entries and tabulated the total uses of all II features in each diary. Before tabulating the entries, authors independently tallied the frequencies of features in three randomly selected diaries (8.1% of the total diaries) and there was 100% agreement in frequency counts for the total number of times each feature was used. The authors then tabulated the features for the remaining diaries for Study 1 (n = 36).
Identifying total number of features used. When tabulating the total frequency of II features, the second and third author also tabulated the use of II features across each diary. Just as participants varied on how frequently they reported using specific features, participants varied in whether they reported using all six functions and all eight attributes. While tabulating the total uses of II features, the second and third author also tabulated how many features each participant used, as well as, their most frequently reported function and most frequently reported attribute, following the procedures outlined in Study 1. There was 100% agreement in these frequency counts in the number of features used in each diary, as well as 100% agreement of the most frequent feature in each diary. The third author then tabulated these data for the remaining diaries for Study 1 (n = 34).
After tabulating the total number of features used, the authors reviewed the diaries from Study 1 to assess the total number of functions and attributes used in each diary. The same coding procedures were used, and similarly, there was 100% agreement in totaling the number of II features across each diary. The authors coded the remaining diaries from Study 1 with this coding procedure. Because of the nature of how frequency and variety were coded in the first study, the total attributes is out of six instead of eight because frequency and variety were not included in the total count for Study 1.
Role Category Questionnaire (RCQ). The Role Category Questionnaire (RCQ) was employed to generate the total number of differentiated interpersonal constructs. Participants completed the RCQ (Crockett, 1965), which asks participants to describe one liked and one disliked peer, during a class session. The second author timed the administration of the instrument, and participants were given five minutes for each section of the RCQ. After the instrument was completed, these free-form responses were coded by the authors, following the procedures in Burleson and Waltman (1988). These responses are coded only to count the total number of differentiations used in the RCQ instrument, creating a representative sample of the participant's construct differentiation. These constructs were not coded for abstractness or integration. The authors coded a sample (n = 4; 11.4%) of the RCQ responses. After an acceptable reliability on this sample was established (Krippendorf's α = .78), the remaining instruments were coded. The scores for participants ranged from 7 to 45 constructs, with a mean score of 26.94 (SD = 8.27).

Results
The second study was undertaken to attempt to replicate the results of Study 1 and enrich the understanding of the associations between II features. Similar to the first study, the diary entries were the primary unit of analysis. There were, totally, 756 entries; 74.4% of the entries (n = 562) qualified as II entries and the remaining analyses focus on these 562 entries. The excluded entries (25.6%; n = 194) reported on general intrapersonal communication (e.g., self-talk, daydreams, and fantasies). Similar to Study 1, we summarize these results in order of the research questions.

RQ1: What function is most reported over extended lengths of time?
In the second study, the function of catharsis was the most commonly reported function; it was present in 20 of the 37 diaries and was represented in 296 total instances across all entries. Compensation was also the least frequently reported function in Study 2. When comparing the observed versus expected frequency distributions for II functions between Study 1 and Study 2, the difference was statistically significant, χ 2 (5) = 92.62, p < .01.
RQ2: What attribute is most reported over extended lengths of time?
The most frequently reported attribute in Study 2 was proactivity (26 of 37 diaries; 367 individual instances), and the least frequently reported attribute in Study 2 was low specificity. When comparing the observed versus expected frequency distributions for II attributes between Study 1 and Study 2, the difference was statistically significant, χ 2 (9) = 72.44, p < .01.

RQ3: Do multiple or singular functions of IIs occur in diary accounts?
The diaries ranged in the number of functions used. There were twelve diaries that used a total of five functions (32.43% of all diaries) and eleven diaries mentioned all six recognized functions in their total entries (29.73% of all diaries). There were nine diaries that used four functions (24.32%), four diaries that used three functions (10.81%) and one diary represented no functions (2.70%). Although not directly part of this third research question, the use of II attributes was also tabulated. For the attributes, 62.2% of the diary entries reflected using five or more of the attributes, while only one diary entry used all eight attributes (2.70%) across the 82-day period.
RQ4: Is there an association between interpersonal construct differentiation and the use of II features over an extended period of time? Our final research question focused on the association between the frequencies of certain II features, as well as. the association between the frequency of certain II features and interpersonal construct differentiation. To answer this research question, a correlational analysis between these three features was performed on these data. Due to the nature of count data, the values were subject to log transformations prior to the correlation analyses. The power to detect significant Pearson product moment correlations between II features with α = .05 based on a total sample of 73 (n = 36 in Study 1 and n = 37 in Study 2) was .21 for small effects (r = .10), .83 for medium effects (r = .30), and in excess of .99 for large effects (r = .50). The power to detect significant Pearson product moment correlations between II features and interpersonal construct differentiation was based on a sample of 37 (Study 2 only) and was .13 for small effects (r = 10), .34 for medium effects (r = .30), and .66 for large effects (r = .50). The correlations between II features and the number of interpersonal constructs appear in Table 1.
The number of interpersonal constructs was positively associated with the total number of entries, the total number of functions used, and the specific II features of catharsis, rehearsal, and proactivity. Additionally, the various associations between II features were also examined in the correlational analysis. A number of intriguing associations are how compensation is related to conflict-linkage (r = .45) and the total number of functions. Hence, compensation allows people to ruminate about conflict, which Honeycutt (2010) describes in his II conflict-linkage theory.
Other notable associations involve conflict-linkage with the total number of functions (r = .64). Conflict is also moderately associated with the frequency of IIs (r = .35) and specificity (r = .45). Retroactivity enhances a deeper understanding of one's values, attitudes, and beliefs (r = .62). However, some correlations were significant, but the associated effect sizes explained approximately one eighth of the variance between constructs. These correlations included the moderate positive associations between interpersonal construct differentiation and total functions, valence and total functions, variety and total attributes, and conflict-linkage and frequency. Additionally, 105 of the 153 total correlation coefficients generated were not statistically supported, which could be due to the sample being underpowered to detect small effects. The total number of entries recorded was positively related to the  total number of features and attributes used, as well as, the specific features of catharsis, rehearsal, relational maintenance, proactivity, retroactivity, and discrepancy. The total number of functions used was positively correlated with the features of conflict-linkage, catharsis, and compensation; the total number of attributes positively correlated to total entries and total functions, as well as, the specific features of compensation, conflict-linkage, frequency, variety, discrepancy, and valence. The largest effect sizes overall were between II features (dominance and specificity, r 2 = .76; proactivity and rehearsal, r 2 = .64). The largest effect size for a correlation involving the number of differentiated constructs was between construct differentiation and the total number of entries (r 2 = .25). Table 1 reports both correlations and effect sizes for all correlations.

Discussion
Both studies presented in this manuscript were undertaken in order to better understand accounts of the frequency of the features of IIs. The first three research questions were duplicated across Study 1 and 2 and as such, will be discussed in more depth in the general discussion section. What Study 2 directly contributes is a deeper understanding of the association between construct differentiation and II features in terms of the frequency of II features.
In this study, high construct differentiation was positively associated with the total number of entries, the total functions used, and the features of catharsis, rehearsal, and proactivity. An individual with more construct differentiation generated more entries and used more functions in these entries. Over the period of time included in this study, more cognitively complex individuals engaged in a wider variety of functions. It may be that less cognitively complex individuals rely on only a few of the functions, engaging in IIs primarily to fulfill only one recurring function, such as planning for an upcoming encounter or releasing tension, whereas more complex individuals are able to engage in IIs to compensate for time spent apart from loved ones or to understand better their personal actions. This potential explanation recognizes the functional communicative differences between individuals with fewer constructs to reflect on or engage and those with a greater differentiated view of social processes. Based on the effect size between the total II entries and construct differentiation, only a quarter of the variance is explained, suggesting that although these variables provide some explanation for the observed differences, there are likely other differences beyond perceiving and recognizing social knowledge constructs explaining our finding.
The most common features (catharsis and proactivity) were positively and moderately associated with construct differentiation. These features were also positively associated with the total number of entries. So, whereas there is a relationship between cognitive abilities and the generation of IIs, there may be other explanatory individual differences, including the use of specific features.
One limitation of study 2 is the statistical power for this sample. Only the larger effects might be seen, as the sample was underpowered to detect small and medium effects. As such, although it is important to confirm these larger, positive associations, we may not have a full understanding of the smaller effects of construct differentiation on the use of IIs. There were no statistically significant negative associations found; the count data used may also be contributing to the findings observed in Study 2. Further limitations of both studies will be discussed in the next section, after a discussion of the shared research questions.

General Discussion
Both studies focused on ascertaining the most common features of IIs over time and how these features occur in various combinations. Study 1 found that rehearsal was the most common function and proactivity was the most common attribute. Study 2 also found proactivity to be the most common attribute, yet found that catharsis was the most common function. Our findings extend earlier studies on message design logic to II theory. Kline and Chatani (2001) have found that higher construct differentiation is associated with greater message effect awareness and that cognitive complexity is related to positive outcomes (Kline, 1991). Specially, proactive IIs are associated with cognitive complexity.
The analysis of interpersonal construct differentiation in relation to II features suggests the need for further advancement of the connections between intrapersonal social cognition and functional communication competence as part of the constructivist perspective. In an explanation of the theoretical background for constructivism and human communication, Delia et al. (1982, p. 163) recognize that "constructs alone do not produce strategies," and II theory may be one way to link these relatively stable knowledge structures, or constructs, with the actions and planning that goes into producing effective functional communication. While planning for upcoming encounters, individuals may rehearse messages that recognize the perspectives of others; similarly, online IIs (occurring during interactions) may allow individuals to formulate high person-centered responses to conversational partners. If IIs do in fact serve as a link between the social construct system and the functional ability to produce competent communication that recognizes the perspectives of others, then continued research in this intersection of social communicative processes could serve to inform both II theory and constructivist theory.
Despite this difference in the most common function, there are some underlying similarities between the functions of catharsis and rehearsal. As Honeycutt (2008) suggests, "when one uses IIs, one experiences anxiety relief" (p. 82). Both catharsis and rehearsal allow individuals to think through possible outcomes of interactions before the actual encounter. In catharsis, thinking through outcomes allows for the release of emotions and anxiety; for rehearsal, thinking through the outcomes prepares individuals with a script for how to engage in the interaction. However, it is important to note that although these features are complementary, they can occur independently or with other features. As seen in Study 1, rehearsal and catharsis occur together, but also occur concurrently with other functions (e.g., catharsis and conflict-linkage; rehearsal and relational maintenance), which are reported by Bodie et al. (2013). Future research could examine the correspondence between II topics in relation to romantic relationships. Our participants were instructed to report on the daily occurrence of IIs regardless of the type of relationship partner they reported on.
Both studies suggest that the functions of compensation and relational maintenance are not as frequently used, providing additional insight into the place of these functions within II theory. Across both studies, the self-driven and self-focused nature of catharsis, rehearsal, and self-understanding may reflect the inward focus of our particular sample: college students. The pressures of finding jobs, planning for the future, or working with other individuals for the first time in group settings may be reasons why this self-focus is prevalent in our findings. This speculation is similar to a conclusion drawn by Bodie et al. (2013), wherein a canonical correlation revealed that compensation was not associated with the other features. Whereas compensation refers to imagined conversations with others one is not actually able to communicate with, the other functions describe uses of IIs to plan for conversations that are likely going to happen at some point in the future or to rehearse conversations that actually did happen at some point in the past.
Whereas individuals did engage in relational maintenance or compensatory IIs with friends and family that could not be seen, it may simply be more common for college-age individuals to use IIs to rehearse, relieve, or relive conversations to develop a greater understanding of the self in relation to others as part of the developmental process of late adolescence. Future research could examine the IIs of people reporting exclusively on romantic relationships and how often conflict, selfunderstanding, catharsis, and relational maintenance is reflected. It is possible that relational maintenance may be taken for granted depending on the quality of the relationship.
It is also possible that it is not only relationship quality or type explaining the low frequency of compensatory and relational maintenance IIs, but also changes in how we contact our intimate partners throughout the day. Nearly thirty years have transpired since the identification of these functions, and during this time there have been changes in how individuals use technology to communicate with loved ones. Although compensatory IIs may still be used when there are no other means for contacting a loved one, individuals may instead be using technology to connect instantly to loved ones, sending a "thinking about you" text instead of engaging in a "thinking about you" II. Similarly, instead of imagining disclosure and relational development, individuals may be sending texts, pictures, or emails to increase intimacy within a relationship. Between the present findings and other recent research, future research testing II theory may wish to focus on compensatory IIs and relational maintenance IIs to inform our understanding and knowledge of these less frequent functions within II theory.
Across both studies, all functions were present in the daily entries. Whereas the distribution of these functions may feature the most frequent functions of rehearsal and catharsis, it may simply be due to general information reported in the various entries. As participants reported entries featuring a variety of partners and a variety of topics, it may be that certain functions are more prominent in particular relationships (e.g., conflict-linkage and roommates or parents; rehearsal and professors or managers; catharsis and friends or family), and this relationship is not clear in the present methodology. Overall, for II theory, this manuscript provides continued support for the six functions (Honeycutt, 2003), but ultimately these findings suggest for II theory that certain combinations of two functions may be more common for multifunctional IIs rather than all six functions occurring simultaneously (e.g., catharsis and rehearsal; rehearsal and relational maintenance; catharsis and conflict-linkage). Future research may wish to test which functions are most likely to occur together when IIs accomplish dual functions.

Limitations, Future Research and Conclusion
The relation between loquacity and cognitive complexity has been the subject of prior academic debate (see Beatty & Payne, 1985;Burleson, Applegate, & Neuwirth, 1981;Burleson, Waltman, & Samter, 1987). Powers, Jordan, and Street (1979) were the first to express concern that the constructs measured with the written RCQ were simply a reflection of the verbosity of the participant, a concern also investigated by Beatty and Payne (1985). Although the full debate cannot be addressed in this space, the measurement of construct differentiation as it relates to loquacity in our data must be addressed.
First, some differences in this lively debate are due to operationalization and measurement of the constructs (e.g., Beatty & Payne, 1985). We followed the recommendations set forth by Burleson and Waltman (1988) in the administration and subsequent coding of the RCQ, in which constructs are typically represented as fragmented ideas varying in length and detail, versus employing measures of cognitive complexity based in sentence completion tasks. Second, our goal of including construct differentiation differs from the general debate on loquacity. Certainly, loquacity may be one explanation for the wide variation in entries we examined. Across both studies, some participants recorded their IIs using only a few sentences, while others filled multiple pages. However, we have attempted to provide some explanation for why individuals differ in their use of II functions, which represents a mental construct and cognitive activity. As such, our investigation of II features in relation to construct differentiation has been undertaken in order to discover how the categorization of interpersonal beliefs and constructs may relate to the use of the functions. General verbal ability and loquacity may be another competing explanation to be explored in subsequent work, and may temper generalizations from our findings. However, we have stayed committed to furthering II theory through this work, and a central tenant of II theory is that imagined conversations are mental representations of conversations and thus, should be studied in conjunction with other mental representations, including constructs and beliefs.
Recalling that IIs are mindful activities (Honeycutt & Bryan, 2011), the occurrence of IIs is certainly not mindless, but the identification of the features of particular IIs may not be as mindful. A limitation of diary studies is that researchers are reliant on the amount of detail and description included in the diary entry. There is wide variation in how participants articulate the features present in their IIs. Some entries rely entirely on implicit statements, where attributes and functions must be inferred, while other entries specifically identify the features present. As such, it may be that IIs include other features, but the imagery associated with the II is fleeting. By the time an entry is recorded, participants may have forgotten what other functions were present, so our findings may only represent those features easily recalled when participants generate the entries analyzed. Additional research should look at the content, as well as the relational context in which the IIs occur.
A limitation of our study was that we could not identify which day of the week that IIs were recorded. Research by  has revealed IIs about families were more likely to occur on weekends. Despite individual differences in frequency, IIs may be a bouting activity. A bouting behavior is defined as an activity that may occur over a time period that is relatively infrequent. Yet when it occurs, the activity occurs in quick, intense bouts. For some people, binge drinking is a bouting behavior-it occurs infrequently, but in intense bouts. Similarly, IIs may occur infrequently, yet after an intense argument, people may be caught in an absorbing state wherein they ruminate about the conflict. Future research could examine whether IIs as a bouting behavior occur in relation to certain days, topics, various seasons, and around events such as anniversaries, dates of passing of loved ones, before job interviews, and family gatherings. Which functions are associated with these bouts? Are people more cognitively complex or less differentiated during these bouts?
Based on these diary entries focused on intrapersonal communication and IIs, there is a high occurrence of cognition designed to prepare an individual for upcoming communication encounters, planning out potential responses or reactions from known partners, or reducing anxiety before interacting with unfamiliar partners. IIs, while representing all the identified functions, tend to focus on single-function or dual-function IIs rather than incorporating multiple functions in a single II. These entries may reflect an element of some IIs in which vignettes can occur rapidly and may form without a specific purpose identified before the II. However, these vignettes may result in understanding the social actor or understanding the function of a particular II. This increased understanding of the frequency of attributes and functions should spark future research opportunities for IIs and intrapersonal communication. Finally, future research should consider the order in which features are recalled reflecting cognitive load capacity. If cognitive load is a potential explanation for the use of multiple features in single IIs, then general cognitive ability may explain why individuals repeatedly use single function IIs or repeatedly use the same II function in their daily lives.
Within our findings investigating the daily accounts of II features, we find continuing support for the occurrence and use of all II features recognized in II theory, but particularly the role of catharsis and rehearsal in preparing for upcoming and anticipated conversations. A key concept in II theory is the link between IIs and conversations. We find continued support that IIs are used in preparation for upcoming conversations by both rehearsing plans for these conversations and relieving our unresolved tensions and anxieties about these conversations. These mindful representations of conversations provide continued support in the role of II theory as an explanatory mechanism linking intrapersonal social cognitive processes with our daily interpersonal conversations.

Notes
[1] Material containing information from this coding scheme on identifying attributes and features along with sample entries is available upon request from the senior author. [2] The sample size for assessing reliability was slightly over 10% of the total diaries (n = 37) available and was selected for numerous reasons. Prior to coding, there was no way to assess how many entries qualified as II focused entries, which is the true focus of this study. Second, despite the fact that much of the work identifying II features was already done by the participants, the diary coding process was time intensive. Selecting a smaller sample size prevented coder drift. [3] Although there were no entries in the inter-coder reliability sample that contained multiple IIs, there were some cases of entries containing more than one II in the coded sample. For these cases (n = 2 participants total, one in each sample; 3 entries each featuring two IIs), the entries were treated as separate entries and the total number of entries for each diary was adjusted to account for the multiple IIs.