Uncertainty reduction theory, set out by Charles Berger and Richard Calabrese in 1975, holds that people meeting for the first time communicate largely to make each other more predictable, and that as uncertainty falls, information seeking such as question asking falls with it. A message thread keeps that process on the page: who asked, in what order, and how the questions thinned out.<\/p>\n\n\n\n
Most accounts of the theory describe two strangers talking face to face, where the questions are gone the moment they are asked. In a chat, every one of them is still there. That makes a message thread an unusually clear place to see what the theory was describing, and an equally clear place to see what it cannot tell you.<\/p>\n\n\n\n
The theory comes from a 1975 article in Human Communication Research<\/em>, \u201cSome explorations in initial interaction and beyond: Toward a developmental theory of interpersonal communication\u201d. A communication theory textbook chapter hosted by Montgomery College<\/a> describes it as the foundation of the theory, which is also known as initial interaction theory, and sums up its central supposition: in a first interaction, a person\u2019s main concern is to decrease uncertainty and increase predictability about the behaviour of both people.<\/p>\n\n\n\n Uncertainty has a specific meaning here. Berger and Calabrese defined it as the \u201cnumber of alternative ways in which each interactant might behave\u201d (p. 100). The more ways a stranger could plausibly act or answer, the higher the uncertainty. The idea was borrowed in part from Shannon and Weaver\u2019s information theory, where uncertainty is high when many outcomes are possible and none is more likely than the others, and in part from Fritz Heider\u2019s work on how people explain each other\u2019s behaviour.<\/p>\n\n\n\n The same chapter notes a limit the authors set themselves. The 1975 article dealt with initial interactions, and Berger and Calabrese did not explore later interaction, though they recommended that future research apply the framework to developed relationships.<\/p>\n\n\n\n Berger and Calabrese tied uncertainty to seven other concepts: verbal output, nonverbal warmth, information seeking, self-disclosure, reciprocity, similarity and liking. Each pairing became an axiom, and from the seven axioms they deduced twenty one theorems. The Montgomery College chapter reproduces the full list in its appendix.<\/p>\n\n\n\n Read against a typed conversation, the axioms split into two groups. Some describe things a written record holds directly. Axiom 1 links the amount of verbal communication between strangers to falling uncertainty, and a thread is nothing but verbal communication, with its volume on display. Axiom 3 says high uncertainty causes more information seeking, and that information seeking decreases as uncertainty declines. Axiom 5 pairs high uncertainty with high rates of reciprocity and low uncertainty with low reciprocity.<\/p>\n\n\n\n Others have no clean equivalent in text. Axiom 2 concerns nonverbal affiliative expressiveness, and a chat has no posture, voice or face. Axioms 6 and 7, on similarity and liking, describe judgements that sit with the people involved rather than in any particular message. A thread can show what was said about a shared interest; it cannot show whether either person felt more alike afterwards.<\/p>\n\n\n\n That split matters for what follows. The parts of the theory that a thread preserves are mostly about behaviour that leaves text behind: talking more, asking, and answering in kind.<\/p>\n\n\n\n Charles Berger later grouped the ways people gather information into three strategies, according to the Montgomery College chapter, which cites his 1987 work: passive, active and interactive. The AURA Lab at Southern Illinois University Edwardsville<\/a> describes them as observing someone unobtrusively, asking third parties or arranging situations, and questioning the person directly.<\/p>\n\n\n\n Only the third strategy happens inside the conversation, so it is the only one a message thread records. Scrolling back through someone\u2019s profile, or asking a friend about them, leaves no trace in the chat itself. A direct question does. That is the first thing worth keeping in mind when reading a thread through this theory: the questions on the page are a part of the information seeking, not all of it.<\/p>\n\n\n\n Research on question asking has added detail that a thread makes easy to look for. William Douglas, in a 1987 study described in the same chapter, examined question asking under different levels of anticipated future interaction and found that mutual question asking was highest when the anticipated future interaction was moderate. A 1990 study by Douglas found that question asking resulted in uncertainty reduction, which in turn resulted in more disclosure.<\/p>\n\n\n\n A 2012 study by M.L. Antheunis, A.P. Schouten, P.M. Valkenburg and J. Peter, published in Communication Research<\/em>, brought the interactive strategy online. According to the abstract on the University of Amsterdam repository<\/a>, 81 unacquainted cross sex pairs were randomly assigned to talk face to face, through a webcam, or in text only. The proportion of question asking was higher in both computer mediated conditions than face to face, and there was no difference in question asking between text only and webcam. That result concerns strangers in a single experimental session, and it is about the share of questions, not their number.<\/p>\n\n\n\n Axiom 5 is the one that most often surprises people. It does not say reciprocity grows as two people get to know each other. It says the reverse: high uncertainty produces high rates of reciprocity, and low uncertainty produces low rates. Theorem 4 follows from it, stating that the amount of communication and the reciprocity rate are inversely related, and theorem 16 states that information seeking and reciprocity rate are positively related.<\/p>\n\n\n\n The chapter does not spell out how reciprocity is measured, so what follows is a reading rather than the authors\u2019 own method. In a thread, the obvious candidates are a question met with a question back, or a detail offered and a detail returned. If the axiom holds, that kind of matching is densest in the early exchanges and loosens as the conversation goes on.<\/p>\n\n\n\n This is the point where a careful reading has to stop short. The axioms describe tendencies across initial interactions in general. A single thread where the questions dry up, or the matching stops, is consistent with uncertainty having fallen, and it is also consistent with a dozen other explanations the text does not hold. The theory gives you something to look for; it does not tell you why it happened in any one conversation. For a broader account of what reading a chat can and cannot establish, see what a chat analysis cannot tell you about someone<\/a>.<\/p>\n\n\n\n A first meeting in person is gone as soon as it ends, and what remains is two people\u2019s memories of it. A message thread is different in three practical ways that bear directly on this theory.<\/p>\n\n\n\n It keeps the order. Axiom 3 is a claim about change over time, information seeking falling as uncertainty falls, and a thread holds the first messages and the most recent ones in sequence. It is possible to put the opening week of a conversation beside a week from months later and compare what each contains.<\/p>\n\n\n\n It keeps who did what. In a spoken conversation it is easy to remember that \u201cwe asked each other a lot\u201d. A thread shows which person asked and which answered, message by message, which is exactly the distinction axiom 5 and theorem 16 turn on.<\/p>\n\n\n\n It keeps the text without the face. That removes axiom 2 from view entirely, which is a real loss, but it also means everything that was available to the two people at the time is still available afterwards. The hyperpersonal model<\/a> takes up why text can build a stronger impression than meeting in person, and grounding in communication<\/a> looks at which properties of a spoken exchange a typed one lacks.<\/p>\n\n\n\n Portraitor is built around the same two choices this reading depends on. On the Portraitor home page<\/a>, the person uploading picks the one person they want a portrait of, and the page says their messages are read while everyone else stays context. The next step asks which time period to analyse, with presets running from all messages to the last month, plus a custom range. How that choice works, and what it leaves out, is covered in analysing only part of a chat<\/a>, and the case of choosing yourself as the person is covered in what picking yourself means<\/a>. The home page does not describe the portrait in terms of uncertainty reduction, and nothing here claims it measures uncertainty.<\/p>\n\n\n\n Berger himself wrote, in 1987, that the theory \u201ccontains some propositions of dubious validity\u201d (p. 40), as quoted in the Montgomery College chapter. The best known challenges go to its centre.<\/p>\n\n\n\n Michael Sunnafrank argued in 1986 that reducing uncertainty is not the primary concern of a first interaction, and that people are more concerned with maximising relational outcomes, an account he called predicted outcome value. Berger replied that outcomes cannot be predicted without some history of interaction, so predicting them is itself a way of reducing uncertainty.<\/p>\n\n\n\n Kathy Kellermann and Rodney Reynolds tested axiom 3 directly. Their 1990 study of over a thousand students found that wanting knowledge was a better indicator of information seeking than lacking it. High uncertainty on its own did not produce information seeking; a wish for the information also had to be there. For a reader of a thread, that is the most useful caution of all, because the axiom a message thread displays best is the one most directly questioned.<\/p>\n\n\n\n Two further limits apply to any reading of a chat through this theory. Berger and Calabrese wrote about strangers, and the chapter notes that Canary and Dainton (2003) questioned whether the theory applies across cultures, since people from cultures with a high tolerance for uncertainty may not find it a primary motivator. And Planalp and Honeycutt showed in 1985 that communication can increase uncertainty when new information conflicts with what someone already knew, so a burst of questions late in a thread need not mean the opposite of what an early burst meant.<\/p>\n\n\n\nSeven axioms, read against a message thread<\/h2>\n\n\n\n
Questions are the visible half of information seeking<\/h2>\n\n\n\n
Matching early, and why the matching fades<\/h2>\n\n\n\n
What a written record keeps that a first meeting loses<\/h2>\n\n\n\n
Where the theory has been pushed back<\/h2>\n\n\n\n