“Silence is a true friend that never betrays” Confucius

Silence in Text-based Computer Mediated Communication

Yoram M Kalman1

October, 2004

This PhD proposal was written under the supervision of professor Sheizaf Rafaeli, at the

University of Haifa Center for the Study of the Information Society

An online version of this document, with active links, is located at:

http://study.haifa.ac.il/~ykalman/Research Proposal.pdf

1 ykalman@study.haifa.ac.ilhttp://study.haifa.ac.il/~ykalman

1

Table of Contents

Summary..........................................................................................................................3Literature Survey.............................................................................................................8

Silence..........................................................................................................................8Online Social Cognition and Chronemics.................................................................12

Social Cognition....................................................................................................12Social Cognition Online........................................................................................13Chronemics............................................................................................................23

Silence in CMC..........................................................................................................25Online Irresponsiveness.........................................................................................27Online Responsiveness and Response Times........................................................30

The Research Question and Its Import..........................................................................32The Research Question..............................................................................................32What is Online Silence?.............................................................................................32

Alternative Definitions..........................................................................................35Intentional and Unintentional Silence.......................................................................35

Intentional Silence.................................................................................................36Unintentional Silence.............................................................................................36

How is Silence Experienced and Interpreted.............................................................37Interpreting Silence................................................................................................37

Conclusion.................................................................................................................39Import of the Research Question...............................................................................40

Contributions to Theory.........................................................................................41Contributions to Practice.......................................................................................41

Methodology..................................................................................................................43Defining Silence........................................................................................................43

Quantitative Definition..........................................................................................43Estimation of Typical Response Rates..................................................................43

Expression and Perception of Silence........................................................................44A Meta-analysis of Research on Online Responsiveness..........................................44Research Populations.................................................................................................44Summary....................................................................................................................45

Phases of the Proposal...................................................................................................46Resources.......................................................................................................................47Appendices....................................................................................................................55

Appendix 1 – definition of “silence” from the Oxford English Dictionary online edition (OED, 1989)..................................................................................................55Appendix 2 – Copy of Kalman & Rafaeli, 2005.......................................................62

2

“Blessed is the man who, having nothing to say, abstains from giving in words evidence of the fact.” George Eliot

Summary

We propose to investigate the way silence is expressed, experienced and interpreted in

text-based computer mediated communication (CMC). Silence has long been a topic of

intensive research in traditional settings, by linguists, communication researchers,

psychologists, scholars of culture and others. Also, the last decade has seen significant

research into the socio-cognitive aspects of human behavior in CMC settings, including

that of lurking (Rafaeli, Ravid, & Soroka, 2004). Nevertheless, very little attention has

been paid to the online manifestations of silence, and to the socio-cognitive aspects of

that behavior.

We wish to build a research infrastructure that will allow the investigation of online

silence. We will start by creating a nomenclature and criteria for defining and measuring

online silence, using quantitative data collected using the unobtrusive measures

approach developed by Webb et al (Webb, Campbell, Schwartz, & Sechrest, 2000). This

work has already started, and its first results will be published in early 2005 (Kalman &

Rafaeli, 2005). In this work we created a responsiveness profile based on the

examination of an unprecedented sample of thousands of email replies. Previous work,

was usually case-based and either probed responsiveness through the reactive method of

personal interviews (Tyler & Tang, 2003) or treated online responsiveness as a side

issue. Here, we were able, through the “windfall” of the confiscation and release of

massive data files of the Enron corporation, to extract detailed behavioral information

without raising privacy and other ethical limitations. The results, which are far more

robust quantitatively than any previous work on online responsiveness, allowed us to

corroborate the findings of Tyler and Tang (2003), extend them, as well as identify a

recurring theme in a variety of datasets, including Enron’s, work on Usenet (Jones,

Ravid, & Rafaeli, 2004), and on Google Answers (Edelman, 2004). This initial work will

be extended by similar investigations, as well as via a meta-analysis of published data

3

about online responsiveness. In addition to chronemics, an attempt will be made to assess

and measure response rates.

Further, we will start investigating the socio-cognitive aspects of online silence by

understanding the ways online silence is created and the reasons it is created. We expect

to focus on an important distinction between two major categories: intentional and

unintentional silence, the former being a result of a decision not to respond, and the

latter being a result of other factors such as a technical mishap, a message missed as a

result of information overload, etc. This distinction is especially pertinent in the case of

online communication, where unintentional silence is believed to be common.

The last socio-cognitive aspect we will investigate is the “receiving end” of silence,

investigating the way online silence is experienced and interpreted. The negative

emotional and physiological effects of online silence have already been documented

(Eisenberger, Lieberman, & Williams, 2003), and we will seek to further understand this

aspect of online silence. In this case, too, we expect the attempt to resolve between

intentional and unintentional silence, to be a key component in the interpretation phase,

since this uncertainty is an inherent element in online communication, acknowledged and

taken into consideration by all participants.

The socio-cognitive part of the research is still in its formative stage. Based on previous

work on similar subjects, we expect to use and apply the following theories and their

tools to the research of the expression, experience and interpretation of online silence:

(1) attribution theory, (2) uncertainty reduction theory, (3) media richness theory and

reduced social cues, (4) theory of media synchronicity, and (5) the SIDE model. These

theories and a first take on their applicability to the study of silence in CMC are reviewed

in the following.

A model of a text-based CMC communicative cycle is presented, a few initial hypotheses

are presented, and a list of proposed research methods is offered. In the proposal, we

discuss in detail the methodological challenges in researching a “non-action”, as well as

the conundrums of ethical and privacy issues linked to researching private online

behavior.

4

We expect that the results of the research described here will provide researchers with a

nomenclature for description and discussion of online silence, as well as a methodology

that will allow measuring and characterizing different forms of online silence in various

text-based CMC media. The empirical work will begin by yielding some benchmark

measures for the chronemics and responsiveness levels of popular forms of CMC. We

also expect this research to test the applicability as well as enrich and extend the above-

mentioned theories in the context of CMC, and to further the understanding of social

exclusion, social translucence, and interactional coherence in relation to silence. We

expect the results of the research to suggest practical policies and technological

suggestions that will improve the quality of online communication, since online silence

breaks online interactivity, and thus hinders CMC.

5

“Silence is a fence to wisdom”Untranslatable Hebrew proverb from the Mishna: Avot 13, 3

Introduction

In this research proposal we suggest to investigate the way silence in expressed,

experienced and interpreted in text-based computer mediated communication (CMC).

In the literature survey, we present the extensive body of research about silence in

traditional settings, emphasizing the multifaceted and ambiguous nature of silence. We

then present a summary of our review on online social cognition (Rafaeli, Raban, &

Kalman, 2004 in press), introducing key concepts in social cognition, and focusing on the

emerging field of research into socio-cognitive aspects of the online behavior of people.

We show that the results of this work enhance our understanding of human behavior both

online and offline, by challenging, enriching and expanding existing socio-cognitive

theories. Next, we examine the role of chronemics and time in interpersonal

communication, outline the literature on response times and responsiveness online, and

review literature that touches on the negative effects of online irresponsiveness,

cyberostracism and online reticence.

Following the literature review, we present the research question: How is silence

expressed, experienced and perceived in text-based CMC? The importance of answering

this question stems from the fact that online silence is a ubiquitous and annoying

disruptor of both personal and organizational communication processes. The answers are

also expected to further our understanding of online socio-cognitive issues. We look at

the constituents of the research question: defining and measuring online silence;

exploring the reasons for the creation of online silence, with a special distinction between

intentional and unintentional silence; understanding the way online silence is experienced

and interpreted, and the way this interpretation might affect the reactions of persons who

experience online silence. We detail a few of our hypotheses in relation to the expression

and experiencing of online silence, as well as present a model of a typical text-based

CMC communication cycle, which illustrates the various stages at which silence can be

6

"שתיקה ,לחוכמה סייג"

created, both intentionally and unintentionally. We acknowledge the fact that researching

silence, a “non-occurrence”, an absence, is complex, as well as somewhat paradoxical.

Methodologically, we describe our intention to take the “unobtrusive measures”

approach (Webb et al., 2000), attempting not only to use non-reactive measurement tools

whenever possible, but also trying to “triangulate” results collected using varied

measurement tools, non-reactive, as well as reactive tools such as interviews and

questionnaires. In order to define silence and establish its prevalence, we suggest

combining a methodology we recently developed (Kalman & Rafaeli, 2005) which

allows defining online silence in various contexts and media, with a meta-analysis of

research on online responsiveness. We then present a suggestion for reactive as well as

non-reactive tools to describe and possibly explain (in light of existing socio-cognitive

theories) the perceptions and interpretations of online silence, as well as the reactions to

it. In all of our methodology we propose practical solutions to the inherent difficulty of

measuring something which is an absence, a response that was not created, a non-event.

In summary, our work proposes to take silence, a phenomenon that has been researched

in traditional settings by various disciplines (linguistics, sociology, communication), and

research its online manifestations. Initial, laboratory based research, has already shown

that the online manifestation of silence can be a powerful phenomenon, and our hope is to

extend this work into the field, and create a methodological infrastructure and a

conceptual framework that will allow the investigation of silence in the era of CMC.

7

“Do not the most moving moments of our lives find us without words?” Marcel Marceau

Literature Survey

Online silence is an online communicative behavior. In this literature survey we present

the current state of research on silence and on social cognition, thus presenting relevant

aspects of three disciplines that provide foundations for our proposed research:

communication, linguistics and social psychology. After presenting the background on

silence and on social cognition, we look more closely at online social cognition, as well

as present the concept of chronemics, a concept from the field of non-verbal

communication, which researches communicative cues related to time. Lastly we present

research that touches, either directly or indirectly, issues related to online responsiveness

and irresponsiveness.

Silence

“Silence is to speech as the white of this paper is to this print” said Bruneau (1973), in an

effort to capture the ubiquity and the centrality of silence in communication, while also

acknowledging that silence is treated by most people as an insignificant background, a

meaningless default, and a useless emptiness. Like other forms of nonverbal

communication, silence is central to communication since how something is said

communicates at least as much as what is being said (Jaworski, 1999). Like the empty

page, which carries little meaning without graphic symbols, and the graphic symbols

which can’t exist without a substrate, so do silence and speech coexist in a mutual

dependence, each one providing the context, and thus the meaning, to the other. In

addition to the speech that surrounds it, silence need also be understood in other contexts,

such as other nonverbal signs as well as in the context of culture (Braithwaite, 1999).

Because it is so context dependent, silence can express and be interpreted as expressing a

wide range of meanings. The scope is so wide that actually silence can, in different

contexts, mean opposites. Jaworski (1999) gives as an example Jensen’s work (Jensen,

1973) where five functions of silence which can have contrasting, positive and negative

values are described:

8

1. A linkage function: Silence may bond two (or more) people or it may separate

them.

2. An affecting function: Silence may heal (over time) or wound.

3. A revelation function: Silence may make something known to a person (self

exploration) or it may hide information from others.

4. A judgmental function: Silence may signal assent and favor or it may signal

dissent and disfavor.

5. An activating function: Silence may signal deep thoughtfulness (work) or it may

signal mental inactivity.

Another example Jaworski gives is the work of Lebra (1987) where on the one hand

reticence is interpreted as a sign of honesty, sincerity and straightforwardness, but on the

other hand it is associated with concealing the truth. Silence amongst young spouses in

the Japanese culture is an expression of affection, while in the same society silence is

used to express social defiance, disagreement with a person, anger and hatred. In

Bruneau’s work (1973) one can find additional examples of contrasting meanings of

silence: as an empowering as well as an oppressive tool, as a promoter of interpersonal

closeness, as well as a sign of aggressive alienation, as a respectful act, as well as a way

to signal disrespect. Silence can signal the end of an interaction with a clear finality, as

well as represent closeness and intimacy. The polite behavior of a person joining an

English group is to join in silence, while the same behavior will be considered

inappropriate and impolite in a Mediterranean country like Greece (Sifianou, 1997). In

educational settings silence can be a sign of active learning and concentration, as well as

of idleness and ignorance (Jaworski, 1999). In an organizational context, silence can be

attributed to lack of motivation and isolationism, while it actually originates in stress and

ambiguous communication (Jenkins, 2000).

The many possible interpretations of silence, compounded by the fact that many of these

possible interpretations are at odds with each other, make silence a topic which is

difficult to categorize and define, and consequently difficult to research. A definition is

elusive and quite futile since “We are likely…to sense the strangeness, frustrations, and

9

ambiguities of silence no matter how we define the concept. Since silence is a sort of

absence of something, it suggests a potential… Concepts of the sort of silence, then, may

be peculiarly difficult if one’s purpose is to objectify the state, that is, to say that it is

something” (Scott, 1993, pg 11). Even language itself is an obstacle to clearly defining

silence, since the English language does not distinguish between two important types of

silence, which are described by at least two words in some languages. The first definition

the Oxford English Dictionary (OED, 1989) (see appendix 1) gives to silence is (1a)

“The fact of abstaining or forbearing from speech or utterance (sometimes with reference

to a particular matter); the state or condition resulting from this; muteness, reticence,

taciturnity”. In German this type of silence would be “schweigen”, and in Hebrew

“shtika”. The second definition provided is (2a) “The state or condition when nothing is

audible; absence of all sound or noise; complete quietness or stillness; noiselessness. In

German this type of silence would be “stille, and in Hebrew “dmama” or “dumia”. Its

closest equivalent in English is “stillness”. Definition 3b which we will later propose to

adopt for the purpose of the present research is “Neglect or omission to write (about

something); failure to communicate or reply”, and would also be translated in German

into “schweigen” and in Hebrew into “shtika”.

Some of the confusion over the meaning of silence stems from the fact that silence has

been researched within the confines of a few different disciplines. Jaworski (1997) looks

at silence in a multidisciplinary manner, as a linguistic, discoursal, literary, social,

cultural, spiritual and meta-communicative phenomena. Each of the disciplines looks at

silence with a different toolset, and the resulting confusion adds to the difficulty of

comprehending this essentially ambiguous subject. Despite the risk of inconsistency

stemming from these difficulties, we would like to present a few classification systems

suggested for categorizing various types of silence. Bruneau (1973) defines three forms

of silence: Psycholinguistic Silence, Interactive Silence and Socio-cultural Silence.

Psycholinguistic Silence, such as pauses which slow down speech while it is being

created, are a result of either the need of the speaker for extra time to perform the

linguistic tasks, or to give the listener time to process the speech effectively. Interactive

Silences, such as the pauses that allow turn-taking in dyads or in small groups, are mutual

silences shared by the members of the dyad or of the small group, until one or more of

10

the parties chooses to break the silence. Socio-cultural Silence, such as the silence during

acts of religious worship, is silence used by entire social and cultural orders in specific

situations. Poyatos (2002) lists and categorizes many forms of silence (and stillness) at

the human level (language, paralanguage, audible kinesics, other body sounds,

direct/indirect acts upon objects/substances), animal, cultural environment and natural

environment. Every form of silence is presented as a silent alternative to a specific sound,

and next to its corollaries of movement and stillness. For example, silent footsteps can

alternatively be noisy, and feet may either move or remain still. Kurzon (1995) looks at

silence in the context of “the right of silence” and analyses it from the point of view of

“intention” and “ability”. A silent response could be a result of the lack of ability to speak

(either due to ignorance or to psychological disabilities such as shyness or

embarrassment) or of the “ability not to speak”, or “ability not to say anything”:

intentional silence. An interesting and diverse list of “types of silence” can be extracted

from the index to Jaworski’s interdisciplinary book (Jaworski, 1997) and consists of the

following types: absolute, acoustic, antecedent, anterior, arbitrary, contemplative,

displayed, gustatory, inter and intra turn, olfactory, spiritual, static, surrogate, tactile,

temporal and visual.

The controversial “spiral of silence” theory was developed by Elisabeth Noelle-Neumann

(1991), claiming that people who hold a minority view are less likely to express it in

public, due to fear of isolation. This reticence results in a spiral resulting in silencing

minority views, a process which is accelerated by mass media.

It is clear that the nomenclature of silence is so extensive and diverse, that it is impossible

to fit “online silence” into one clear category. Silence in CMC can be intentional or

unintentional; it can be psycholinguistic, interactive or socio-cultural; it can be arbitrary

contemplative, visual, and so on. In this work, we suggest to focus on one paradigmatic

type of silence, the one which is represented in definition 3b from the Oxford English

Dictionary (OED, 1989): “Neglect or omission to write (about something); failure to

communicate or reply”. This definition covers situations which we find as the most

commonplace and intuitive examples of online silence or stillness: an unanswered email

message, a posting to an online forum (e.g. chat forum, Usenet, online classroom) which

11

does not receive a response, or an unanswered instant message. It leaves for later research

forms of online silence such as large white spaces between paragraphs in online

messages, empty postings, the silence in online synchronous audiovisual events, empty

spaces in web pages, outdated and unavailable links, etc.

Online Social Cognition and Chronemics

Social Cognition2

Social cognition (SC) deals with the mutual influences of cognition on social life, and of

social environments and processes on cognition. Cognition is fundamentally influenced

by the social environment (Levine & Resnick, 1993). Research on social facilitation,

social loafing, social roles, and mental representations has shown distinct social

influences on cognitive abilities and task performance. SC is about the cognitive

underpinnings of social behavior (Devine, Hamilton, & Ostrom, 1994). SC studies how

social structures and social processes are mentally represented, and how social interaction

is important for the development and practice of cognition. Individuals are viewed as

being engaged most of the time in information processing. Information is encoded from

a social context, is interpreted, elaborated, evaluated, inferred and attributed. Processed

information, or knowledge, is later used in judgment processes and for guiding behavior.

Research into SC shows that judgment and behavior need not always be the result of a

thorough mental process, but rather the result of short-cuts known as heuristics. SC draws

from both social and cognitive psychology. It deals with how people make sense of

themselves and of others. Some refer to social structures as concepts or schema (Kunda,

1999). One key theory related to the way social inferences are made is attribution theory

(Heider, 1958), asserting that people constantly answer a need to make sense of the world

they live in by attributing traits to others. The traits are attributed by drawing inferences

from the behaviors of others. Attribution is a three-step process: Perception of the action,

judgment of intention, and, finally, attribution of disposition (Griffin, 2003). Behaviors

can, for example, be attributed to internal or external factors, to controllable or

uncontrollable factors, and to stable or unstable causes. The attribution is based on

antecedents such as prior information about the event and its circumstances, beliefs, and

2 This section is adapted from (Rafaeli et al., 2004 in press)

12

motivations. The results of the attribution can influence not only behavior, but also

produce emotional or affective reactions, as well as alter expectations about self and

about others (DeJoy, 1994). An interesting angle on attribution theory looks at the early

stage of personal relationships, and emphasizes that attempting to explain the behavior of

others at these early stages is fraught with uncertainty. Berger’s uncertainty reduction

theory (URT) (Berger & Calabrese, 1975) proposes eight axioms to explain the

connection between uncertainty and eight variables: verbal communication, nonverbal

warmth, information seeking, self disclosure, reciprocity, similarity, liking and shared

networks (Griffin, 2003).

The basic social structures mentioned in the literature describe person traits or perception

(Fiske, 1993). These include attitudes, beliefs, stereotypes, implicit personality theory,

and salience. The main social processes in SC research have been attribution, attitude

change, impression formation, social comparison, decision making, social construction

of reality or joint sense making (Fulk, 1993).

Social cognition is not limited to the study of individual cognition and how it is affected

by the social environment. SC impacts the way in which individuals and cultures

perceive, define, and interpret media in general, and in our special case, the Internet.

Social Cognition Online3

Research on SC among individuals, dyads, and groups has focused on behavior and

perception in physical environment (Fiske, 1993; Fiske & Taylor, 1991). When turning

to the virtual, it is tempting to set up a contrast with the “natural” way that people

interact, i.e. face-to-face (Chapanis, 1975; Lantz, 2001). The question then is, of course,

if face-to-face should be treated as the benchmark against which all innovations and new

contexts need be judged? One prime example of elevating face-to-face traditional set-ups

to ideal/standard status is the media richness approach (Daft & Lengel, 1986). Media

richness theory argues that task performance will be improved when capabilities of the

media (cues, feedback, personal focus, and language variety) are matched to task

ambivalence and uncertainty. In this approach face-to-face communication is considered

3 This section is adapted from (Rafaeli et al., 2004 in press)

13

the richest communication medium in a hierarchy followed by the telephone, electronic

mail, letter, note, memo, special report, and finally, flier and bulletin. Some observations

and predictions are made regarding the propriety and efficiency of different media.

Specifically, this theory suggests that performance in equivocal tasks would be better

when using “rich” media. According to this theory, in the case of unequivocal tasks,

performance would be better if leaner media are used. Unfortunately, empirical data to

support media richness theory fall somewhat short (Dennis, Kinney, & Hung, 1999;

Dennis & Valacich, 1999; El-Shinnawy & Markus, 1997; Morris & Ogan, 1996). An

interesting theory which developed from a critical examination of media richness theory

is the theory of media synchronicity (Dennis & Valacich, 1999) which proposes that

media choice is influenced by five media capabilities (feedback, symbol variety,

parallelism, rehearsability, reprocessability) to support two fundamental communication

processes (conveyance and convergence).

It is important to note that face-to-face interactions are neither an ideal nor should be

treated as an ultimate standard. SC develops in a variety of loci/media. Previous research

has already contested the standard of face-to-face asserting that interpersonal interactions

and social influences affect media choice (Fulk, Schmitz, & Steinfield, 1990), and online

relationships have been shown to be healthy, and a complement to face-to-face

relationships (Peris et al., 2002), and based on unique information seeking strategies

(Ramirez, Walther, Burgoon, & Sunnafrank, 2002). With the Internet there is the

possibility that face-to-face be demoted from its ostensibly classic preordained

position/status as ultimate yardstick. The Internet itself is a plurality of media operated by

diverse technologies which constitute a culture or a social space in its own right. In fact,

the “richness” of CMC is a variable, not a characteristic.

The CMC social space is characterized variously. Slater (2002) proposes four properties

which describe the online environment as a space in its own right:

Virtuality – a computerized representation of reality. Spatiality – the network is a space

in its own, not parallel to real life space, and cannot be mapped onto offline spaces.

Disembedding – a community can be spatially dispersed. Geography has all but lost its

14

meaning to participants of online communities. Disembodiment – most Internet

communication is textual and often anonymous. Slater's framework stresses the fact that

people do not have to reveal any of the offline cues used for SC such as age, sex, race and

location. The online identity can be very different than the offline identity.

Impression formation in CMC

The perception of the “other” online and how the impression of those we interact with

online is formed has been a focus of research from the first days of online

communications. The initial focus was on the gap between the impressions formed

online, and impressions in “real life”, with an emphasis on anecdotal cases of fraud and

deception. This approach follows a historical pattern of focusing on the sensational, as

well as on the faults and deficiencies of a new medium in relation to traditional

communication media. Similar claims were made historically with the introduction of

early electric media (Standage, 1998) and in social science research about the first days

of telephone and telegraph (Pool, 1983). Much of the early work focused on the reduced

social cues in comparison to face-to-face communication. The reduced social cues

approach highlighted the surprise and disappointment that arose when those who formed

the impressions were confronted with “real life”. These works concluded that online

impression formation is faulty and wrought with stereotypical and prejudiced

assumptions used to “fill in the blanks” (Albright, 2001) of the reduced social cues, and

terms such as “fluid identities” (Turkle, 1997) were used to warn about the unsound and

shifting sands of CMC.

In retrospect it is clear that many of the early works on CMC may have failed to

distinguish between the various contexts of online activities. Some popular activities such

as MOOs and MUD were purposely structured for “play” purposes, wherein

impersonation and identity experimentation were the expressed purpose of these settings.

Such environments flourished on university campuses among students, close to the eye

and attention of researchers. This, too, may have given these contexts some increased

salience. No wonder that a “reality check” in such cases reveals that the (generally)

young and often experimentally minded people behind the screen names are different

than imagined. Reaching general conclusions about human behavior online based on

15

these environments is somewhat similar to generalizing about human character after

observing actors perform a play on stage.

As indicated above, impersonation, (a form of self-presentation), ranges from simply

using an alias name in order to save typing or time or avoid revealing one's real name

through assuming an identity for playful purposes, and, finally, to creating a complete

false identity in order to engage in criminal or terrorist activity. The latter form was

given the most media attention but the other two forms of self-presentation are by far

more widely used. All these forms of self-presentations indicate a high degree of

awareness of others, your own and others’ perception, combined with a high degree of

understanding of the computerized and networked environment.

Research shows that senders often try to optimize their self-presentation by mentioning

information they perceive as impressive, while holding back information which is less so

(Walther & Burgoon, 1992). The receiver in the online case sometimes idealizes the

sender, “filling in the blanks” with information that tends to be too rosy (Tidwell &

Walther, 2002). Such a combination can result in disillusion when eventually a face-to-

face meeting occurs. This disenchantment received much attention in literature dealing

with online dating and online relationships (Turkle, 1997), but is apparent in other online

contexts too (Rouse & Haas, 2003). As Rouse and Haas point out, inaccuracies in

personality perception of online “others” are mainly a result of three important

differences between Internet-mediated factors and face-to-face communication: The first

is that physical appearance has a less meaningful effect, the second is that people may

behave different online than in a face-to-face situation, and third, that online there is a

heightened level of ambiguity due to the lack of vocal inflection and facial expression. In

contrast, it is important to point out that such assertions about the inferiority of online vs.

offline impression formation ignore some more basic questions. For example, is the

ability to present an idealized self online similar to the ability to idealize one's physical

appearance through the choice of clothing, haircut, makeup, accessories and even plastic

surgery or over polite behavior? Are humans gradually becoming more skilled at

detecting signs of such online attempts, integrating them into the emerging impression,

just as they would detect and interpret an attempt to conceal physical imperfections in

16

traditional offline interaction? Is the ambiguity of an online smiley or online silence

analogous to some extent to the different interpretation a smile or silence can have in an

Eastern culture like Japan in comparison to its meaning in a Western country like

England? Are some people better at putting on an online “poker face”, while others'

“online face” is easily interpreted by competent users? Is the reaction to a slick and

overly polite used cars salesperson offering us a once in a lifetime deal if we make our

minds immediately, before someone else grabs it, similar to the reaction to “spam”

promising us the opportunity to make thousands of dollars a month working from home

in our spare time? We contend it may be too early in the evolution of human online

communication to simply conclude that the ability to discern authenticity online is

inferior to face to face. Rather, it is important to focus on understanding the dynamics

and evolution of the emergence of the human capability to communicate online, on the

way skilled users are using it to fine-tune their perceptions, or to influence the way they

are perceived. Reeves and Nass (2000) point out in their discussion of the “perceptual

bandwidth” of computer mediated communication: “… the assumption that more is

always less is misguided. An increase in the breadth and depth of media representations

certainly turns up the volume knob on perceptual responses, but greater presence does not

translate into greater efficacy or desirability; intensity does not equal quality.”

Once the issue of the superiority of face-to-face over CMC is removed, the questions that

arise are questions that focus on aspects of SC online, and mainly questions of what

influences the way users translate the special social cues of CMC (especially text-based

CMC) into impressions, what influences these impressions, and how do CMC users try to

influence the way they are perceived by others. Online impression formation occurs in

stages. First impressions are formed based on very initial signs such as the email

addresses or the screen names as well as on the context of the online occurrence

(Wallace, 1999). As the interaction progresses, more information accumulates: the

sentence structure, vocabulary, grammar, spelling, use of capital letters, typographical

marks such as emoticons, as well as self-testimonials about matters such as gender, age,

location, occupation, hobbies and marital status. Additional elements that can influence

this impression are less linked to pure linguistic aspects and can include the length of the

message, the amount devoted to talking about oneself, number of opinions expressed,

17

level of friendliness, the content of the message, strength of expressed opinions, (Savicki,

Kelley, & Oesterreich, 1999), as well as non-verbal cues such as response time (Tyler &

Tang, 2003; Walther & Tidwell, 1995).

How do all of these cues translate into an impression of the “other” online? As in offline

situations, this process is in essence inductive, and may be based on social stereotypes,

categorizing people based on the signals they give off. Success of this process depends on

factors such as the context of the interaction, the interpreter’s capabilities and information

sources, the self presentation of the “other” and his or her willingness to present

truthfully. Cognitive psychology has developed a keen interest in the introspective

notion of heuristics. How do people form impressions, perceptions, cognitions and

attitudes by relying on shortcuts. Prospect theory (Kahneman & Tversky, 1979) explains

a wide variety of perceptual and attitudinal peculiarities of human choice, preference and

behavior. Much of prospect theory has been developed in traditional contexts. How does

prospect theory translate to the online arena? It seems that, at least in the case of one

effect, the endowment effect (Thaler, 1980), the predictions that applied to the physical,

tangential world work well for the abstract, virtual and online milieu. In their experiments

(Raban & Rafaeli, 2004; Rafaeli & Raban, 2003) for instance, found that people apply

similar decision making biases and heuristics in online contexts. Consistency in behavior

overpowers opportunity for change. Such attempts to understand the online manifestation

of established theories (such as socio-cognitive theories) are imperative to both

advancing human behavior online, as well as to expanding the scope and hopefully

validity of the theories.

One of the theories dealing with the way these scarce cues are translated into impressions

is the Social Identity model of Deindividuation Effects (SIDE) developed by Spears and

Lea (1994), theorizing that these limited cues undergo an “over-attribution” process,

which may also result in a stereotypical assignment of attributes to group members

(Chenault, 1998). Even few social cues can make a significant impact on the quality of

impressions people form of each other, and it is through interpersonal and “intimate”

contact that people are able to cross the boundaries of ethnocentrism and stereotypes

(Tanis & Postmes, 2003).

18

Perception of self online

In addition to understanding how others are perceived online, it is also important to

understand how people perceive themselves when online. Self perception or self concept

may seem fixed or constant. In fact, self perception is very malleable. SC research has

identified a variety of selves: Self concept varies in private versus public circumstances,

it varies with the different roles we assume and contexts or situations we experience.

Thus, self-concept is dynamic with attention focused on the context-specific self rather

than on one 'global' self (Devine et al., 1994). Sherry Turkle in a seminal article (Turkle,

1980) provided computers and networks’ impact on perception of self with a slogan that

captures the variability of self-perception, perhaps amplified, by computers and networks.

In "Computers as Rorscharch" she says that users project meaning to computerized

activities rather than being passive recipients. This sentiment was then echoed in much

HCI work on user-centered computing (Shneiderman, 1998), and naturally percolated

into the very design of systems and networks. The projected meaning is influenced by a

variety of environmental or social effects, in addition to the personality of the actor.

Further research in Turkle's tradition evolved in the landmark book "The Second Self"

and in "Life on the Screen" (Turkle, 1984, 1997). The Internet provides numerous

opportunities to affect what Goffman would call “the presentation of self” (Donath, 1998;

Stone, 1991). The construction of a personal home page, the introduction one is expected

to make when entering an online forum, the short descriptions many provide as a rite of

inclusion into various social software arenas, and the profiles one accumulates for oneself

willingly or not on a variety of online systems, all influence presentation of self. In a

work on response time to email messages, the researchers (Tyler & Tang, 2003) found

evidence that email users used response time as a tool to influence the way they present

their “responsiveness profile”.

Groups online

An additional area of interest in online impression formation is the way people perceive

themselves as members of online groups. What are online groups? Many terms have been

used to describe influential Internet based interactions between several participants,

including “virtual communities” (Rheingold, 1993), “virtual teams” (Lipnack & Stamps,

19

2000) and “virtual groups” (Wallace, 1999). Here, we will use the more generic term,

virtual groups. The tools used by people to congregate online are diverse, and include

email-list forums, synchronous chat systems such as IRC, asynchronous discussion

forums, MUDs and MOOs, Usenet newsgroups, virtual classrooms, web logs, and

groupware tools. Some of these are very rich media, allowing real-time transmission of

audio, video and text, as well as online application sharing, while others are very

rudimentary, and based only on the asynchronous transmission of ASCII text. People

participate in online groups for work, education and leisure, acquiring, disseminating and

sharing information and knowledge, collaborating, and socializing.

Initially, there was a lot of excitement about the possibilities opened by online virtual

groups, with the combination of a widely dispersed but closely-knit community. A good

example is The Well, established in 1985 and described in the book The Virtual

Community (Rheingold, 1993). At about the same time, virtual communities were

contrasted with “real” communities, while mourning the damage inflicted by

“Technopoly” (Postman, 1992). Later the term “Internet Paradox” was coined (Kraut et

al., 1998), claiming that the Internet actually reduces the level of social involvement of its

users. An additional element in this process of disillusion was the accumulation of online

communities that disintegrated, or simply stopped functioning. Unlike “real world” social

connections which can disappear without leaving a trace, online communities often leave

behind artifacts (Jones, 1997) which can confuse people who might attempt to join or

interact with such communities, only to discover that they are totally silent (Wallace,

1999). The disillusion was supported by reports of discussions in online groups that

resulted in flaming and polarization (Spears, Russell, & Lee, 1990). When put into

perspective such a dichotomous debate becomes meaningless (Etzioni & Etzioni, 1999;

Wellman & Gulia, 1999). Online groups are as “real” as any other groups, and are simply

different in some aspects from traditional groups. Online groups are social units in which

the participants are interdependent, and behave according to explicit or implicit social

norms. Online groups show, just like other groups, both examples of social compensation

as well as social loafing, and even effects such as crowding and deindividuation (Spears,

Postmes, Lea, & Wolbert, 2002) have been documented.

20

“Lurking” is a special form of SC in groups. Participation in online groups, such as forum

discussions, is not symmetric. Usually, a small number of participants contribute. A much

larger number remain receivers. Often, reticent receivers are named “lurkers” (Nonnecke

& Preece, 2000). Receivers who remain passive are either intimating a social cue – or at

the very least are understood as imparting a message. Thus, lurking is not just a behavior.

It is a perception and is perceived by others in social contexts. Lurking has been a social

and cognitive concern since the early years of public CMC. Even in participatory virtual

communities, many people limit their participation to reading and never post themselves.

The reported proportion of lurkers varies from around 90% (Katz, 2003; Mason, 1999) to

around 50% (Nonnecke & Preece, 2000; Soroka, Jacovi, & Ur, 2003). Lurking usually

means "lying in wait", often with malicious intent. But interestingly enough, “lurking”

does not have to carry a negative connotation. The Merriam-Webster dictionary

definition of the verb "lurk" has one unexpected meaning – "to persist in staying". Thus

lurkers can be defined as a persistent but silent audience. Lurkers have been recognized

by many researchers as an important and integral part of any community. Rafaeli and

Sudweeks (1997) point out that though lurkers are an important part of any online group

there is little information about their activity. Whittaker et al. (Whittaker, Terveen, Hill,

& Cherny, 1998) also acknowledge lurking as a very popular activity among virtual

community participants that leaves no traces.

The reasons for lurking range from concerns for privacy, through respect for others’ time

and attention limits, to those rooted in personality (Rafaeli et al., 2004). In any case, both

the reasons for lurking and its outcomes are central to the understanding of SC among the

less salient and available – but probably more numerous -- participants in online social

behavior.

Online social presence

Social presence is the awareness of others’ being there. As early as 1976, Short et al.

(Short & Christie, 1976) identified social presence as crucial to the understanding of

mediated behavior. Presence was postulated to affect trust, compliance, attraction,

motivation, and more. More recent work on virtual presence looks at virtual presence as

both an independent and a dependent variable. As computerized systems and networks

21

are designed by humans, the degree to which they elicit a sense of presence is an

important variable. Thus, for instance, Biocca and Levy (1995) experiment with the

interaction effect between user factors and media factors on feelings of social presence .

They provide strong evidence for human's automatic social responses to artificial

representations possessing humanistic properties such as language and personality.

Clearly the more immersive CMC systems become, with a wider range of sensory and

cognitive appeals and a longer average exposure people have to these systems, the degree

to which such systems are actually social in the full sense of the word comes into focus.

Lombard and Ditton (1997) remind us that virtual presence might be an illusion or a

hoax, and that its effects are worthy of further examination.

Interactivity

One central driving force in the induction of computers into our social circles is the

notion of interactivity. Interactivity refers to the extent to which communication reflects

back on itself, feeds on and responds to the past. Interactivity is the degree of mutuality

and reciprocation present in a communication setting. The term interactivity is widely

used to refer to the way content expresses contact, and communication evolves into

community. And, of course, interactivity is a major option in governing the relation

between humans and computers (Rafaeli, 1984; Rafaeli, 1988; Rafaeli, 2004).

Interactivity is an essential characteristic of effective online communication, and has an

important role in keeping message threads and their authors together. Interactive

communication (online as well as in more traditional settings) is engaging, and loss of

interactivity will result in a breakdown of the communicative process. Research of

rhythms in email and other CMC media resulted in claims that text-only CMC is

“Interactionally Incoherent”: disjointed, without clear turns, and in general “chaotic”.

But, as noted by Herring (1999), text-only CMC is extremely popular, despite obstacles

such as disrupted turn adjacency and lack of simultaneous feedback. The online

interaction is highly desired, and almost addictive in nature (Caplan, 2003; Morahan-

Martin & Schumacher, 2000).

22

Chronemics4

Time and social interaction

A key non-verbal cue affecting online impression formation and communication is

Chronemics: time related messages, such as those conveyed by timestamps (Liu, Ginther,

& Zelhart, 2001; Walther & Tidwell, 1995). Chronemics research studies the role of time

in communication, and the importance of chronemics in interpersonal communication is a

direct result of the key role of time, and specifically of temporal rhythms as underlying

organizing factors in social life in general, and in communication in particular. As

Scheflen points out (1982) the discovery of interactional rhythms was a part of the

maturation of the understanding of human communication. Communicators need to act in

synchrony, and synchronization is based on timing and sequencing. Feldstein (1982)

focuses on the importance of temporal patterning in face to face impression formation,

where cues such as speech rates, tempo, pauses and the frequency of talking turns

influence and reflect the way the participants in the conversation perceive each other. A

key concept in understanding the interplay between time and social behavior in

“Entrainment”, defined by Bluedorn (2002) as “… the process in which the rhythms

displayed by two or more phenomena become synchronized, with one of the rhythms

often being more powerful or dominant and capturing the rhythm of the other.” Probably

the most powerful entraining mechanism we know is the light and darkness cycle, which

affects natural rhythms such as sleep, as well as social rhythms such as working hours.

The strength of this powerful entraining principle is evident when these natural rhythms

are forced to resynchronize following rapid travel across time zones – commonly known

as “jet lag”. Examples of strong biological entraining rhythms are the menstrual cycle and

circadian rhythms. Through entrainment, people learn what rhythms to expect: they learn

that finding an academic in the office is much less likely in August than it is in October,

and that finding a parking place in a commercial zone is far easier at 7 am than it is four

hours later. Moreover, people learn that divergence from these patterns is a sign that

something may be wrong: if an acknowledgement to the receipt of an academic

manuscript to a journal is expected to be a few weeks, once a month has passed, the

researcher will probably inquire what went wrong at the journal office. Another example 4 Adapted from (Kalman & Rafaeli, 2005)

23

given by Bluedorn is the expected time to receive a reply on an email. An excessively

quick response (within seconds of delivery) is usually a sign of something gone wrong:

an “undeliverable” response or an auto-reply.

Online chronemics

Chronemics also plays an important role in online communications, and is clearly an

example of an online non-verbal communicative cue (Walther & Tidwell, 1995). Despite

the initial surprise that a “reduced cues” medium such as text based CMC can convey

subtle “non-verbal” cues, it is clear that chronemics are a part of CMC, and

understanding the role of chronemics is key to understanding CMC (Lane, 2004). For

example, it is a central part in the explanation to the apparent paradox raised by Herring

(1999) regarding the interactional incoherence of CMC (see Interactivity section above).

From a chronemics point of view, text-only CMC allows interactional rhythms not

available in other communicative forms, such as multiple simultaneous exchanges.

Moreover, text-only CMC allows people to achieve “hyperpersonal communication”

“… that is more socially desirable than we tend to experience in parallel face-to-face

interaction.” (Walther, 1996). One of the key principles behind this enhanced social

desirability is that when the qualities of CMC are put to good use, CMC allows people to

“disentrain” their own rhythms from those of others, without losing the ability to continue

effective communication. Thus, people are able to keep their activities synchronized,

without the common requirement that the activities be also simultaneous. An important

example of a corporate-wide effort to enhance the disentraining elements of email

communication is the “Yourtime” initiative in Intel corporation (ITsharenet.org, 2004).

We argue that Chronemics should play a central role in enhancing the translucence of

systems. Efforts to increase the social translucence (Erickson & Kellogg, 2000) of CMC

are still exploratory in nature (Lockerd, 2002), and raise technological, social and ethical

dilemmas (Zweig & Webster, 2002). However, Chronemic measures such as response

time in CMC, its expression, mapping and interpretation are among the more promising

and perhaps least threatening of the potential translucence cues.

24

Silence in CMC

One can almost say that research has been too silent about silence in CMC. Nevertheless,

a significant amount of work published about CMC touches on issues related to silence

and irresponsiveness, or at least on the issues of response time to online messages. These

will be reviewed in the next two sections, under the framework of a simple model of a

typical text-based CMC communication cycle. This model, presented in Diagram 1,

simplifies a dyadic text-based online communicative cycle such as email into seven

stages. Each stage can result in a continuation of the cycle, or in its disruption. In the

latter case, the result will usually be silence. This model echoes the

communication/persuasion model developed by McGuire (1981).

25

Diagram 1

Typical text-based CMC communication cycle

26

Yes

Not now

4) Message read

5) Inferred a response expectation

2) Message Arrived

3) MessageOpened

7) Create response/message

1) Send message

6) Decision to respond

No

Not at all

Recipient wants other party to know the decision not to respondRecipient does not want other party to know the decision not to respondRecipient is indifferent to what the other party knows about the decision not to respond

No

No

No

No

No

In our work we will elaborate the stages of this cycle, as well as analyze the various

occurrences that can occur at each of the stages. For example, in stage 1 email silence can

be caused by: a writer who writes a reply but decides not to send it because s/he is not

happy with the result; by a writer who writes a reply but wants to take time to consider it,

or have a later look at it for possible modifications; by a writer who thinks s/he has sent

the email, but in fact it is left in the outbox and does not go out (for example due to a

mistake in offline email synchronization); by a writer who mistakenly places the email in

the draft folder and does not notice the mistake; by a fault in the server of the sender

which results in no release of the email; and by a fault in routing to the intended recipient.

Stage 6 is unique for two reasons. It is a fork with three, rather than two, possible

outcomes. The two disruptive options are then subdivided into additional categories,

based on the aims of the person making the decision not to respond, whether s/he wishes

the (intentional) decision to be perceived as intentional or as unintentional.

This model can also be applied, possibly with some adaptation, to other online media. For

example, in synchronous chat, stage 1 silence can be the result of a message written but

without hitting “enter” (intentionally or unintentionally), of mistakenly erasing the

message while sending it away, and not noticing that an empty line was transmitted, or of

a failure of the sender’s computer, or of the computer connection to the Internet, just

before the sending.

It is important to note that at each of the stages, four factors may be involved: the user,

the client, the server, and the network. Each of these factors is involved in the various

stages, and the multiplicity of factors involved, as well as the various interactions

between them, add to the inherent uncertainty of online silence.

Online Irresponsiveness

Online silence and irresponsiveness have already been implied in theories and studies of

online interactivity (Rafaeli, 1988; Rafaeli & Sudweeks, 1997). Anecdotal evidence to

the need to acknowledge silence as a factor in human-computer communication is nicely

27

described by Nicholas Negroponte from MIT’s Media Lab in work that was carried out

already in 1978 (Negroponte, 1994). A key area of specific research on irresponsiveness,

already described above (See Groups online section above), is the research of “lurking”

(Nonnecke & Preece, 2000; Rafaeli et al., 2004). Obviously, lurking is a special form of

online silence: usually people post into forums expecting to receive a response to their

messages. Lurkers who read and do not respond are, in essence, silent, but do they fulfill

the definition of silence we chose for the purpose of this work: Neglect or omission to

write (about something); failure to communicate or reply? On the one hand, they are

silent and do not respond. On the other hand, it is obviously expected that many of the

readers will not post a response to each and every message posted to the group, otherwise

threads will become endless, and the number of posted messages will swell

exponentially, especially as the number of participants increases. In addition, lurking

focuses on people who never take active part in the discussion, while silence focuses on

specific instances of irresponsiveness. Thus, lurking is a manifestation of silence in a

group situation, similar in many respects to a face-to-face situation involving an

observing audience, with varying degrees of audience involvement. The silence looked

into in the work described in this proposal is more focused (though not exclusively) on

dyadic relationships.

Cramton (2001) documented the important disruptive effect silence can have on teams

attempting to collaborate online. The work shows that the difficulty of interpreting online

silence leads to uncertainty in some cases, and to wrong conclusions in other cases:

“Over the course of the project, it became clear that silence had meant all of the

following at one time or another: I agree. I strongly disagree. I am indifferent. I am out of

town. I am having technical problems. I do not know how to address this sensitive issue. I

am busy with other things. I did not notice your question. I did not realize that you

wanted a response”. This finding reflects the same difficulties and ambiguity linked to

silence in traditional settings, as described before (see section Silence above). Difficulty

interpreting the meaning of silence was identified in Cramton’s work as one of the most

common problems, negatively affecting 100% of the researched teams, and the reasons

included misinterpreting silence as consent when it stemmed from disagreement or

28

inattention, and silence due to technical problems or faulty information misinterpreted as

intentional nonparticipation.

Early research on the “spiral of silence” in CMC has been carried out by McDevitt et al.

(McDevitt, Kiousis, & Wahl-Jorgensen, 2003). The work was performed in the context of

chat room discussions of an abortion related scenario, and did not reveal evidence to

support the model in this specific setting.

Several tools have been devised to assist people who are interested in tracking the

progress of their sent email, the main purpose of these tools is to remove some of the

uncertainty inherent in this form of communication. If we look at a typical text-based

CMC communication cycle such as the one depicted in Diagram 1, we can map which

areas of this cycle the tools relate to. As an example, we will use two email server

packages: MS-Outlook (Microsoft, 2003), and First Class (OpenText, 2004) and one

email tracking service: Readnotify.com (Readnotify.com, 2004).

Stage MS-Outlook First Class ReadNotify.com1 No Yes Yes2 Yes No No3 Yes Yes Yes4 No No Length of time open5 No No No6 No No No, but will report when reopened7 No Yes No, but will report when reopened

Table 1: email tracking tools and the stages in the communication cycle (Diagram 1) covered by them

It is interesting to see how different mail packages provide varying information, and how

a tool like ReadNotify.com tries to overcome some of the deficiencies of existing email

packages. We plan to conduct a further survey of such tools and their capabilities, as well

as identify areas not covered by existing tools. The systematic mapping of such "blind

spots" of existing technology can assist in identifying developmental needs of future

CMC tools, and improve their sociability and usability (Preece, 2000).

29

Online Responsiveness and Response Times

Other than the above mentioned work on online irresponsiveness, there is a relatively

extensive body of work on online responsiveness, and response times, which also

touches, usually as a matter of marginal interest, on the issue of no response at all,

namely of silence. This work is focused in a few areas: responsiveness and response time

to customers who email an organization (Customer-Respect-Group, 2004; Hirsh, 2002;

Mattila & Mount, 2003; Stellin, 2003; Strauss & Hill, 2001), responses to online surveys

(Lewis, Thompson, Wuensch, Grossnickle, & Cope, in press; Sheehan & McMillan,

1999), a few works on responsiveness to business correspondence (Abbott et al., 2002;

Pitkin & Burmeister, 2002; Tyler & Tang, 2003), and work on response times in

discussions on Usenet (Jones et al., 2004) and to questions posted to the “Google

Answers” website (Edelman, 2004).

A superficial examination of these reports reveals a recurring theme, which was

corroborated by us in detailed work carried out on the responsiveness profile of Enron

employees using email (Kalman & Rafaeli, 2005). You may find a copy of this yet

unpublished work in Appendix 2, and on http://study.haifa.ac.il/~ykalman/PID37522.pdf.

In this work we created a responsiveness profile based on the examination of an

unprecedented sample of thousands of email replies. Previous work, was usually case-

based and either probed responsiveness through the reactive method of personal

interviews (Tyler & Tang, 2003) or treated online responsiveness as a side issue. Here,

we were able, through the “windfall” of the confiscation and release of massive data files

of the Enron corporation, to extract detailed behavioral information without raising

privacy and other ethical limitations. The results, which are far more robust quantitatively

than any previous work on online responsiveness, allowed us to corroborate the findings

of Tyler and Tang (2003), extend them, as well as identify a recurring theme in a variety

of datasets, including Enron’s, work on Usenet (Jones et al., 2004), and on Google

Answers (Edelman, 2004). The recurring theme seems to be a concentration of most of

the responses within a relatively short period of time (usually averaging around one day

in asynchronous media), and a spread of ever increasing response times at a relatively

very low frequency. If the frequency of responses is plotted against the duration until the

30

response, the resulting distribution is highly skewed to the left, with a stretched out and

rapidly diminishing right tail. This distribution was best described, in the Enron dataset,

by a Gamma distribution. We intend to generalize these findings and further corroborate

them by performing a meta-analysis of research reporting response times in CMC, as well

as by additional unobtrusive extraction and analysis of additional responsiveness profiles.

This work comprises the first part of our research question.

31

“The world would be happier if men had the same capacity to be silent that they have to speak”Baruch Spinoza

The Research Question and Its Import

The Research Question

The research question we wish to explore in the dissertation is: How is silence

expressed, experienced and perceived in text-based CMC?

We will start by defining what online silence is. Following that, we will explore the

expression of online silence and try to understand what causes online silence,

distinguishing between the intentional and unintentional creation of online silence.

Lastly, we will explore the receiving end of silence by investigating what is the

experience of silence, and what influences the way this online silence is perceived and

interpreted. We will seek to understand the similarities and differences between the way

silence is experienced and interpreted online, and in traditional, offline settings.

What is Online Silence?

The definition of a concept which embodies uncertainty, whose meaning varies in

different contexts and cultures, and which has been subjected to various disciplines of

research, is inherently difficult. This difficulty is compounded by the fact that silence is

defined by the absence of something (a response, a communication), a lack which can

only be defined once the expected response does not happen. Many attempts have been

made at defining silence in traditional settings, and the definition we have suggested as

the relevant one to CMC is definition 3b from the Oxford English Dictionary: “Neglect or

omission to write (about something); failure to communicate or reply” (OED, 1989). But,

this is not an operational definition and is subject to personal interpretation. An

operational definition should be able to characterize, in a context sensitive manner, the

period of time of irresponsiveness which is long enough to be defined as silence. We

suggest defining it statistically, by measuring typical response times in a specific context,

and by defining silence as a period of no response longer than even very long responses.

We suggest that a silent period longer than 99% of the response times in this specific

32

context will be defined as silence. For example, in the context of our Enron data, not

receiving a response within 20 days will be defined as online silence.

This definition is still, to a certain extent, subjective, since the aggregation of response

times in a “specific context” is subjective, and the setting of a limit at 99% is an arbitrary

limit. Nevertheless, it is an operational definition which can be used, and we might try to

confirm that it is in line with the intuitions of users of CMC, i.e. that at times longer than

this “lower limit” people no longer expect to receive an answer, and that at times shorter

than this lower limit, people still think there is a reasonable chance to receive a response.

The ability to generalize from this definition is subject to a wider analysis of

responsiveness profiles in various situations of online message-response pairs, an

analysis which we hope to perform through a meta-analysis of publications reporting on

response times in online communication.

The “lower limit” will vary between different contexts, and it is important to try and

define what are the major variables influencing the lower limit, and in what manner they

influence it. It is impossible at this stage to define a-priori which of the factors will be

investigated by us, and will prove to be important factors, but the following factors are

likely to be important:

Medium synchronicity: we hypothesize that in synchronous media (such as IM),

response times will be significantly shorter than in asynchronous media (such as email or

Usenet discussion boards) simply since in synchronous communication the participants

usually sit by the computer at the same time, and are expected to read incoming messages

as they come, and to respond. Failure to respond actually means the end of the

conversation. Nevertheless, the border between asynchronous and synchronous media is

blurring, with people using purportedly asynchronous email for rapid exchanges of email

(Kalman & Rafaeli, 2005; Tyler & Tang, 2003), and not responding immediately to

synchronous chat and instant messaging which can take place in a few parallel sessions,

as well as while multi-tasking (Shiu & Lenhart, 2004).

Content of message: we expect that different messages will result in different response

times. For example, messages that require to perform a specific action before responding

33

(such as retrieving a piece of information) will show a longer response time than

messages which do not require any actions before responding. Responses that require

careful wording will take more time than those that do not. Urgent and high priority

messages might receive attention earlier, and thus be responded to earlier. In addition, we

expect to be able to identify cues and signs that signal the wish to receive a response, or

to receive is in a speedy manner. Nevertheless, we would also expect responses to

messages which do not specifically ask for a response, and there are also differences in

the degree of “questionability” of questions in messages, for example: rhetoric questions.

Frequency of message checking: Since messages obviously have to be read before a

response is created, frequency of checking for new messages will inevitably influence

response time. This is correct not only for asynchronous media, but even for written

synchronous tools, which do allow people to wait a while before reading incoming

messages.

Sender recipient relations: We expect the identity of the sender and of the recipient to

influence response time. An obvious example is spam: people who receive unsolicited

commercial email are quite unlikely to respond to it. Other examples could be

hierarchical relations within a workplace, client/customer relations, teacher/student, etc.

In addition, it is expected that the identity of others to whom the message is sent (for

example through “carbon copying”) will influence the responsiveness, as has already

been demonstrated in a work about “diffusion of responsibility” online (Barron &

Yechiam, 2002).

Cultural and psychological background of recipient: as with other non verbal

communication cues, we expect chronemics to differ between different cultures, for

example such that can be distinguished as separate “speech communities” (Hymes,

1974). Moreover, we expect different personalities to have different responsiveness

profiles, based on their attention to detail, tendency to procrastinate, etc.

The ability to contextualize response times and take into consideration the influence of

the last four categories on responsiveness behavior will be limited at the stage of the

research covered in this proposal. Before being able to contextualize, we will attempt to

34

generalize, define, and construct a methodology for measurement of the key variables

involved in any type of online silence, and create the foundations and benchmarks that

will then allow a closer look at specific contexts such as cultural, personal and situational.

Alternative Definitions

An additional way to look at online silence is to look for the written equivalent of

“pauses” in speech. If, for example, we look at email, some texts are written as one block,

with hardly any pause signs such as punctuation marks or new lines, while others use

punctuation marks, and paragraphs for clearer structure. These might signal rapid, pause-

less speech, and patient measured speech, respectively. In this respect, it might be

important to understand the issue of turn taking cues. We have chosen not to look into

this aspect of silence, and focus on the more paradigmatic message-response pair,

following the time-honored tradition of interactivity research (Rafaeli, 1988).

Another alternative definition of silence might be a subjective one, defining silence as the

time by which the recipient decides there is no longer a realistic chance to receive a reply,

or, on the replier’s side, when a conscious decision not to reply is taken. This last

definition touches, of course, only on advertent silence. This alternative subjective

definition is not conducive to the quantitative and unobtrusive approach chosen by us for

its objectivity and testability, though it is important to point out that in some particular

cases such a subjective definition is the best way to contextualize the communication

going on.

Intentional and Unintentional Silence

After defining online silence, we will explore the “silent side” of the equation, in an

attempt to understand the reasons for creating online silence. The treatment of the side

producing the silence will focus on the difference between intentional silence, defined as

a conscious decision not to reply, and unintentional silence, which covers all other

possibilities. Unlike traditional communication methods, unintentional silence has a more

prominent role in text-only CMC. Looking at Diagram 1 which describes a typical text-

based CMC communication cycle (for example in email), it is possible to see that

unintentional silence can be a result of many events: A message written but mistakenly

35

left unsent (stage 1); A message sent into cyberspace but not arrived (stage 2); A message

arrived but left unopened or opened unsuccessfully (stage 3); A message opened but left

partially or fully unread (stage 4); A message read but not understood to require a

response (stage 5); A message whose recipient intended to reply on later, but failed to do

so (stage 6). A message whose recipient intended to reply immediately, but did not (stage

7). Thus, all of the “no arrows”, lead to online silence, with the exception of the stage 6

“not now” option, which might or might not lead to silence.

Special treatment will be given (especially in the context of asynchronous media) on a

hybrid state, in which a conscious decision is made to postpone a reply (stage 6 “not

now” option), a conscious decision which can result in three outcomes: a later conscious

decision to reply or not to reply, or an unconscious decision not to reply (forgetting to

respond, and/or running out of time). In this discussion of the important role of

unintentional silence in text-only CMC, we will be able to use the nomenclature devised

by Kurzon (1998). As a background reference for this discussion we may attempt to

estimate what percentage of questions are answered in given media and contexts, for

example, what percentage of emails in which a question is asked, actually receive a reply.

Intentional Silence

We will collect information about intentional silence, possibly as an “impression

formation” tool to influence “responsiveness image”, an expression of status, a way to

assert power, a polite way to convey a difficult message, a tool for avoidance, as well as a

mechanism for indecision and procrastination. An interesting complexity will be added

by the “evasive answer”: a response which might look like an answer to the query, but is

actually not an answer at all, but rather a response that does not answer the question. We

hypothesize that one of the key factors taken into consideration in cases when a conscious

decision to not respond is taken, will be the level of certainty the sender of the original

message has about whether the silence is intentional or unintentional.

Unintentional Silence

We will collect information on unintentional silence. Possibly as a result of

communication overload, of ineffective inbox management (in email and other

36

asynchronous media), of parallel processing (communicating while doing other actions

such as other communication, work, etc) esp. in synchronous media, of not understanding

that a response is required (possibly as a result of missing a part of the message, or

misinterpreting turn taking and closure cues), as a result of procrastination, as well as a

result of technical failure. Specific attention will need to be given to the emerging role of

false positive spam filtering of emails in the loss of email messages. We hypothesize that

two key factors in unintentional silence in email usage will have to do with

communication overload, sometimes in conjunction with ineffective mail folder

management.

How is Silence Experienced and Interpreted

The “receiving end” of the silence also needs to be researched, and the questions will be

how is the silence interpreted, and what actions are taken subsequently. The discussion of

the interpretation phase will be in the context of attribution theory (see Social Cognition

section above), looking at the perception of silence, judgment of intention (or

unintentional silence), the attribution of disposition, the antecedents of the attribution, as

well as the results of the attribution: behavioral, emotional/affective, and altered

expectations about self and others.

Interpreting Silence

Possible interpretations of silence

The possible interpretations of silence will follow the same structure as defined in the

previous chapter and Diagram 1, since the interpretations can be of intentional or

unintentional silence created at all seven stages. It will be interesting to try and

understand how this interpretation process is structured, and we hypothesize that a key

element in this process will be uncertainty, since uncertainty has a dual role in this

process. The first is the significantly greater uncertainty about CMC silence when it

comes to being able to discriminate between intentional and unintentional silence. This

uncertainty makes it more difficult to unequivocally attribute the silence to an intention

of the silent party. The second role of uncertainty, in the case of intentional silence, is the

uncertainty inherent to the interpretation of silence as a non-verbal cue, an uncertainty not

37

unique to CMC. We hypothesize that the interpretation process will, at least in its initial

phases, focus on minimizing uncertainty both about the “intentionality” of the silence, as

well as about its interpretation in case it is intentional. We expect to link this uncertainty

to Berger’s uncertainty reduction theory axioms, and especially to axioms 1, 2 and 5

which are particularly pertinent in the case of online silence (see section Social Cognition

above).

The effects of various interpretations

Different interpretations of the silence will result in different decision about who is

responsible for the silence, and consequently in different decisions on responses to the

silence and to different affective results. According to attribution theory, the result is a

function not only of the interpretations and attribution, but also of the personality of the

decision maker, and other factors.

Responsibility

The three main categories of responsible parties are the sender, the recipient and third

parties. For example, the sender can take responsibility for the silence if s/he finds out the

message was never sent, or that it might not have been clear from the text that a response

is expected. The sender can also put the responsibility on the recipient’s shoulders, for

example assuming the recipient understood that a response is expected, but decided not to

answer, or that the recipient does not have good email habits and thus did not find the

message, or decided to reply on it at a later time and forgot about it. Lastly, the sender

can assign the responsibility to third parties, such as technology (the message got lost in

cyberspace, was classified as junk mail and discarded, was in an unreadable format) or

others (a personal assistant who filters emails, a child who inadvertently erased

something from a parent’s inbox).

Affective results of different interpretations

The unpleasant experience of online silence has been described before (Rintel & Pittam,

1997; Williams, Cheung, & Choi, 2000; Zadro, Williams, & Richardson, 2004). They

report increasing frustration and hostility, and Williams et al. coined the term

38

“cyberostracism” to describe the negative experience of being ignored online. The effects

of cyberostracism have even been documented neurologically by performing a functional

magnetic resonance imaging (fMRI), suggesting that the social pain elicited by online

exclusion is analogous in its neurocognitive function to physical pain (Eisenberger et al.,

2003). Different interpretations of the silence can result in different affective results: the

results might be negative feelings, anger at or frustration with the responsible party. The

strength and direction of the feelings need to be explored.

Responses

The combination of the above (interpretation, responsibility and affective consequences)

can result in different responses to the silence. People might “probe” the recipient by

sending another message, moving to a different media (including face-to-face) or

resending the message. They might also decide not to do anything, and they might alter

their behavior the next time they interact with that one person (ask again in a different

manner, retaliate by being silent or taking long to respond).

Conclusion

The process of creating and interpreting online silence is expected to be a socio-cognitive

act that includes iterative cycles of impression formation as well as attribution and

uncertainty reduction. We expect uncertainty to be one of the major elements involved in

these cycles, and that its role will be significantly greater than in traditional “face-to-face

silence”. We expect uncertainty to be more influential due to the reduced social cues of

the relatively poor mediums used for text-only CMC, and we expect a significant element

of these iterative cycles to focus on the ability to distinguish between intentional and

unintentional silence. We will attempt to show that the main difference between

“traditional” silence and online silence is the relative ubiquity of unintentional silence.

We hypothesize that an initial stage in interpreting online silence is forming an opinion if

the silence is unintentional (leading to one subset of reactions) or intentional (leading to

another subset of reactions to silence). The importance of this ability to distinguish

between intentional and unintentional silence will influence not only the interpretational

phase, but also the phase of silence creation, possibly both in attempts not to create

39

unintentional silence, as well as in attempts to increase the ambiguity of intentional

silence.

Table 2: Summary of key initial hypotheses mentioned in this proposal. Please see

Phases of this Proposal below:

Topic Initial hypothesisDistribution pattern of responsiveness profiles

The shape of responsiveness profiles will be similar to the one identified in the Enron research (Kalman & Rafaeli, 2005), i.e. highly skewed to the left, with a stretched out and rapidly diminishing right tail, possibly best described by the Gamma distribution.The x axis of the above distribution will be a function of the synchronicity of the medium, and of the frequency of checking for new messages. That frequency will also affect the width of the left hand peak: the less often messages are checked for, the wider it will be.

Creation of intentional silence

A key factor in the decision not to respond and to create intentional silence will be the ability of the original sender to know if the silence is intentional or unintentional. Commercial email (“spam”) will probably be an exceptional case.

Creation of unintentional silence

Two key factors in the unintentional creation of silence will be communication overload as well as, in he case of email, ineffective inbox management.

Interpreting online silence

A key element in the initial phase of interpreting online silence will be an effort to minimize uncertainty as to whether the silence is intentional or unintentional.

Cyberostracism A significant proportion of the cases of negative reaction to online silence will stem from unintentional silence.

Import of the Research Question

The importance of silence in communication is well established, and is reflected in the

many expressions idioms and quotations associated with it, as well as its omnipresence in

various arts. Text-based CMC is an emerging form of communication which is at present

a key communication channel for interpersonal and organizational exchanges (Wellman

& Haythornwaite, 2002) . It has already been established that the appearance of new

communicative technologies has more than superficial effects, and may have major social

impacts, and even restructure consciousness (Ong, 1988). Thus, the research of online

silence has both theoretical as well as practical importance.

40

Contributions to Theory

We expect the results of this work to contribute to the theory and application of

interactional coherence issues in CMC. We expect not only to use attribution theory to

further understand online silence, but also to contribute to the extension of this theory

into the online arena, as well as provide examples of its applicability in highly ambiguous

contexts. We believe the results will add a new angle to the issue of media poverty and

thus influence the theories of media richness and of reduced social cues, as well as the

theory of media synchronicity in the context of text-only CMC. We expect uncertainty

reduction theory to be enriched by knowledge gained about the nature of early and

introductory online interactions. We expect our results to support or to weaken the SIDE

model. Lastly, we hope the results will contribute to the understanding of issues in social

exclusion and of issues of social translucence in online communication.

Contributions to Practice

Since text-based CMC is such a ubiquitous communication tool, and since very strong

negative feelings have been linked to online silence, it is our belief that it is in the public

interest to be better informed about this non-verbal cue. For example, 61% of 750

European office workers polled by palmOne (CNN, 2004; PalmOne, 2004) said business

decisions are being delayed due to lack of email response. One quarter of the respondents

said they have to chase for a response to more than half of all the emails they send (70%

in Italy), and 11% of the British respondents have pretended not to have received an

email when chased. We hypothesize that in many of the situations of online silence which

lead to strong negative feelings, there was no calculated intention to create this strong

effect. Unlike traditional face-to-face cues, which result can be experienced immediately,

many users of tools such as email are not aware of the effect advertent or inadvertent

silence can have, and for them findings about online silence will be surprising and

interesting. In addition, we believe that there is a strong curiosity in the public as to how

common intentional and unintentional silences are, what can be done to distinguish

between the two, and how to handle each. The knowledge gained from our research can

educate the public and through that lower the overall level of uncertainty linked to the use

of text-based CMC. We hope to be able to provide, as a supplement to the academic

41

content, a list of such stress reducing tactics, tools and habits that are based on our

research findings, such as the initial list proposed by Collett and palmOne (PalmOne,

2004). For example, a well-known tip is that sticking to a single question per email

increases the likelihood or receiving a response. We hope not only to supplement the list

with additional ideas, but also to substantiate them.

42

“Well-timed silence hath more eloquence than speech”M T Tupper

Methodology

Defining Silence

Quantitative Definition

Silence will be defined using responsiveness profiles of several text-based CMC media,

in a given context, as described by us (Kalman & Rafaeli, 2005). We will attempt to build

the responsiveness profiles in as unobtrusive a manner as possible through methods such

as anonymous lurking, and post-factum collection of response times from available

archives. The main reason to use the “unobtrusive measures” approach (Webb et al.,

2000) is the added validity gained through the use of nonreactive measures, as well as

from triangulations of results from various experiments. In addition, we will analyze

published data about response times in CMC using the same methodology, and if

required augment the non-obtrusive information through questionnaires and interviews of

CMC users. We will use the responsiveness profiles to define the 99% mark, the “lower

limit” as described above, of various media and contexts.

Estimation of Typical Response Rates

In an effort to estimate what percentage of the questions asked in text-based CMC receive

an answer (response rates), we will attempt to nonreactively identify questions, and

measure the rate of response to these questions, as well as question users of CMC. For

example, using the Enron database, we will attempt to analyze a number of users who

have a reasonably sized “sent items” folder by manually identifying emails which asked

for an answer, and searching the full database for an email responding to that question.

This method has several limits, since the answer might not have been captured in the

database, might have been given using a different medium such as phone or face-to-face,

or might simply not be identified by us. In addition, not all responses are a result of a

clear question asked in a previous message. Nevertheless, this information can be

“triangulated” with more obtrusive (reactive) information received through interviews

and questionnaires of CMC users. Through such reactive methods, we will also try to

43

identify the parameters that can contextualize specific cases, so as to provide not only a

general benchmark, but also tools to move from these general averages, to concrete

situations.

Expression and Perception of Silence

A combination of interviews and questionnaires will be used to assess reasons and means

of expressing online silence, as well as ways online silence is perceived, attributed, and

reacted on. The questions will focus on various factors hypothesized to influence the

expression and perception of online silence, for example such as those used in research

on the role of various theories: SIDE (social identity model of deindividuation effects),

SIP (social information processing), and URT (uncertainty reduction theory) in the initial

stages of CMC interactions (Tidwell & Walther, 2002) or of attribution theory in specific

contexts (DeJoy, 1994). Special attention will be given to the difference between

intentional and unintentional silence, using methodologies such as those used to research

intentional and unintentional embarrassment as well as other types of social rejection

(Keltner & Buswell, 1997; Sharkey, Kim, & Diggs, 2001; Twenge, Catanese, &

Baumeister, 2003). We will also try to confirm the findings suggested by these reactive

methods, by looking for unobtrusive evidence in archives such as the Enron corpus. It is

also possible that we will use techniques such as interaction process analysis (Bales,

1950) as well as other qualitative tools (Orthmann, 2000).

A Meta-analysis of Research on Online Responsiveness

A meta-analysis will be performed on published work relating to online responsiveness.

This meta-analysis will cover both peer-reviewed academic publications which report on

online responsiveness such as Sheehan & McMillan (1999), as well as industry reports

such as those published by customerrespect.com (Customer-Respect-Group, 2004). See

section online responsiveness and response times above.

Research Populations

We plan to reach as diverse populations as possible, since silence is a ubiquitous

phenomenon across all online user populations, and since our intention is to try and reach

44

general results that can then be contextualized in specific cases. We assume that the

populations we will reach will be using text-only CMC for professional, recreational and

educational purposes. We hope to reach demographically diverse populations, since

clearly different demographic groups display different online behavior (Rice, Shepherd,

Katz, & Dutton, 2004 in preparation). Privacy concerns will always limit our ability to

reach research populations in an unreactive manner, but a “windfall” opportunity like the

Enron Corpus (Kalman & Rafaeli, 2005; Klimt & Yang, 2004) can assist in overcoming

these challenges.

Summary

This proposal covers a wide spectrum of methods, both quantitative and qualitative, all of

which aim to create a methodological basis for the in-depth investigation of various

aspects of online silence. Some of the methodologies already exist, and others will need

to be developed, by adapting existing methodologies and measurement tools. Table 3

summarizes these methodologies.

Table 3: methodologies

Purpose Methodology/tool Based onDefine online silence

Responsiveness profiles of various audience and media combinations, based on archives, logs and other databases

(Kalman & Rafaeli, 2005)

Meta-analysis of published data about online responsiveness and irresponsiveness

Estimate typical response rates

Unobtrusively measure from databases

(Webb et al., 2000)

Interviews and questionnairesUnderstand expression and perception of silence

Interviews and questionnaires Methodologies employed in the following areas of research such as: attribution theory, uncertainty reduction theory, and social information processing.

45

“The silence, often of pure innocence, persuades where speaking fails”William Shakespeare

Phases of the Proposal

This proposal suggests a multi-stage research plan. Initially, we suggest continuing

research we started in the last year, which attempts to create a clear definition of online

silence in text-based CMC. We have relatively well formed, though not yet formally

stated, hypotheses that need to be established or refuted, as well as a methodology to

create the definition, and access to databases and populations necessary for this ground

laying phase. Following the results of the first phase, we propose to explore the socio-

cognitive basis of the behaviors associated with expressing, experiencing and interpreting

online silence. The hypotheses and methodologies of this later phase are less formed than

those of the first phase, and will, to a large extent stem from these findings. Despite the

relative imprecision and vagueness of that follow-up stage, we believe it is feasible and

executable, based on similar socio-cognitive work presented above, which looked at the

online manifestation of human behavior. We acknowledge that at this phase we are

unable to describe our plans in detail. We believe that the explanatory success of

attribution theory and of the uncertainty reduction theory, as well as the interesting

findings on lurking, interactional coherence, and social translucence provide a firm

enough theoretical basis to substantiate the assumption that online silence will be

amenable to research with similar tools, and will result in valuable findings.

Despite the fact that our intended purpose is to research online silence and

irresponsiveness, a significant part of the work will have to focus on understanding online

responsiveness, and then derive the data about irresponsiveness and silence from its

complement. This limitation is a result of the surprising scarcity of data about online

responsiveness, a paucity highlighted by our perspective on irresponsiveness. This need

to produce extensive information about responsiveness, and not only about silence, will

allow us to come up with an additional deliverable in conjunction with the last phases of

the research: a list of suggested policies and technological improvements aimed at

improving online communication, responsiveness, and responsiveness perception.

46

“It is difficult to keep quiet if you have nothing to do.”

Arthur Schopenhauer

Resources

47

Appendices

Appendix 1 – definition of “silence” from the Oxford English Dictionary online edition (OED, 1989)

silence, n.

1. a. The fact of abstaining or forbearing from speech or utterance (sometimes with reference to a particular matter); the state or condition resulting from this; muteness, reticence, taciturnity. Occas. with a or in pl.    a1225 Ancr. R. 78 Ine silence & ine hope schal beon ower strenc e. Ibid., Heo mei ec hopien et heo schal ec singen urh hire scilence sweteliche ine heouene. c1375 Sc. Leg. Saints xxxiii. (George) 637 [He] gert scilence be mad, til he had sad at wes in his gule. 1388 WYCLIF Acts xix. 33 And Alisaundre axide with his hoond silence, and wolde elde a resoun to the puple. c1420 LYDG. Assembly of Gods 44 In Plutoys name [there was] commaundyd silence. 1474 CAXTON Chesse II. iii. (1883) 38 And oftetymes they selle as welle theyr scilence as theyr vtterance. 1535 COVERDALE Job xxix. 21 Vnto me men gaue eare,..& with sylence they taried for my councell. 1597 SHAKES. 2 Hen. IV, II. ii. 178 No word to your Master that I am yet in Towne. There's for your silence. 1601 CAMPION Wks. (1909) 24 Doe not demaund why I am mute: Loues silence doth all speech confute. a1668 DAVENANT News fr. Plymouth I. i, Silence becomes Men best, when Women talk. 1720 OZELL Vertot's Rom. Rep. I. III. 157 The People intimidated, kept in a profound Silence for some Time. 1781 COWPER Retirement 414 The tongue..Shall own itself a stamm'rer in that cause, Or plead its silence as its best applause. 1818 SHELLEY Julian 363 Nor dream that I will join the vulgar cry; Or with my silence sanction tyranny. 1847 EMERSON Poems Wks. (Bohn) I. 471 Ye taught my lips a single speech, And a thousand silences. 1875 FARRAR Silence & Voices ii. 29 What is called the silence of ignorance may sometimes be the silence of repudiation, sometimes even the reticence of scorn. 1967 G. STEINER Lang. & Silence 415 Dickens, Hopkins, Kipling are examples of modern writers whose root sensibility was oral, and who tried to adapt essentially oral means to the silences of print. 

   personif. 1607 SHAKES. Cor. II. i. 192 My gracious silence, hayle: Would'st thou haue laugh'd, had I come Coffin'd home? 1667 MILTON P.L. IV. 604 The wakeful Nightingale..all night long her amorous descant sung; Silence was pleas'd. 1815 SHELLEY Alastor 65 And Silence, too enamoured of that voice, Locks its mute music in her rugged cell. a1875 G. M. HOPKINS Poems (1967) 31 Elected Silence, sing to me. 

48

    b. In the phrases to keep (or hold) silence, to break silence, in silence.    (a) a1225 Ancr. R. 22 Vrom et, efter Preciosa, holde silence. c1290 S. Eng. Leg. I. 228/324 Hov holde e so silence at neuer on ne spekez with o ur? c1375 Sc. Leg. Saints xl. (Ninian) 336 As ai come til his presence, he gert e puple kepe scilence. c1450 Rule Syon Monast. liii. in Collect. Topogr. I. (1834) 31 Alle schal there kepe hyghe silence. 1471 CAXTON Recuyell (Sommer) I. 122 With this Iupiter helde his pees and kept scilence. 1560 J. DAUS tr. Sleidane's Comm. 18 He required him that his adversaries might kepe silence. 1782 PRIESTLEY Corrupt. Chr. II. IX. 211 They kept a strict silence all the week. 1819 SHELLEY Cenci IV. iv. 88 He keeps firm silence; but these lines found on him May speak. 1841 LANE Arab. Nts. I. 102 This is an event respecting which it is impossible to keep silence. 

   (b) 1390 GOWER Conf. I. 86 Thus fulofte my silence I breke. c1400 Destr. Troy 2525 Than Troilus..brake Sylense belyue, and abrode saide. 1590 SPENSER F.Q. I. i. 42 He [Morpheus] mumbled soft, but would not all his silence breake. 1667 MILTON P.L. IX. 895 At length First to him~self he inward silence broke. 1718 POPE Iliad XIX. 461 He broke Eternal silence, and portentous spoke. 1783 BURNS Poor Mailie 12 At length poor Mailie silence brak. 1842 BROWNING Pied Piper iv, An hour they sate in council, At length the Mayor broke silence. 

   (c) c1380 WYCLIF Sel. Wks. I. 93 We shulden be tymes reste, and preye to God in scilence. c1430 LYDG. Min. Poems (Percy Soc.) 41 He kept the nyhte in peas and silence. c1450 St. Cuthbert (Surtees) 993 an sole in silence sall he sitt, And rays him self abouen his witt. 1610 HOLLAND Camden's Brit. (1637) 566 In speech will I ever render thankes, and in silence acknowledge my selfe most deepely endebted. 1746 FRANCIS tr. Horace, Epist. I. xvii. 75 But had the Crow his Food in Silence eat, Less had his Quarrels been. 1757 W. WILKIE Epigoniad VII. 192 Amaz'd we stood; in silence, each his mind To fear and hope alternately resign'd. 1827 in Scott Chron. Canongate Introd. App., The next toast..he wished to be drunk in solemn silence. 1889 Sat. Rev. 9 Feb. 145/2 A brave man suffers in silence. 

    c. to put to silence, to silence by argument or prohibition; to put to death; also to put silence (un)to, to reduce to silence.    (a) 1382 WYCLIF Matt. xxii. 34 Pharisees, heerynge that he hadde put silence to Saducees. 1508 KENNEDIE Flyting w. Dunbar 41 Heir I put sylence to the in all partis. 1677 A. YARRANTON Eng. Improv. 155, I know writing Books of Trade..puts a silence unto the whole History, be it never so good.

49

 

   (b) 1502 ARNOLDE Chron. (1811) p. xxxvii, The ii. sonnys of Kinge Edward were put to silence. 1529 MORE Dyaloge I. Wks. 127/1 Ye haue put me to sylence, that I dare not nowe bee bolde to tell you that I haue sene it my selfe. 1579 W. WILKINSON Confut. Fam. Love Brief Descr. iiijb, Which wordes so often he repeated, that thereby he put Barry to silence. 1601 SHAKES. Jul. C. I. ii. 290 Murrellus and Flauius, for pulling scarffes off Cæsars Images, are put to silence. c1680 BEVERIDGE Serm. (1729) I. 499 So as to put them to silence. 1846 TRENCH Mirac. xix. (1862) 326 He had put them to silence and to shame before all the people. 1879 M. J. GUEST Lect. Hist. Eng. xlii. 424 Tyndale..would..sometimes put all the dignitaries to silence by his arguments. 

   fig. 1581 J. BELL Haddon's Answ. Osor. 254 Such force and dexterity, as may be able to putte your overthwart obstinacy to scilence. 1590 J. SMYTH in Lett. Lit. Men (Camden) 60 My little Booke..shall be put to silence and abolished. 

    d. Used imperatively, = Be silent; make no noise.    1590 SHAKES. Mids. N. V. i. 266 But silence, heere comes Thisby. 1667 MILTON P.L. VII. 216 Silence, ye troubl'd waves, and thou Deep, peace. 1728 POPE Dunc. III. 165 Silence, ye Wolves! while Ralph to Cynthia howls. 1819 SHELLEY Cyclops 475 Silence now! Ye know the close device. 1873 SYMONDS Grk. Poets vii. 225 Silence! Hush! what noise was this? 

    e. The renunciation of speech chosen or vowed by certain religious or monastic orders, esp. the Trappists; a period during which the members of a community or retreat renounce speech. Freq. in phr. the rule of silence.    1387, c1450 [see sense 7]. a1631 DONNE Poems (1633) 69 Harmelesse fish monastique silence keepe. 1884 ADDIS & ARNOLD Cath. Dict. 804/1 Probably the most trying part of all the discipline is the silence, no monk being allowed to speak to his brother on any occasion. 1921 G. O'DONOVAN Vocations xxii. 305 Hush, Sister. The rule of silence is no joke. 1957 P. L. FERMOR Time to keep Silence 67 There is a special dispensation from the rule of silence for the monks who deal with the abbey livestock when they are actually addressing their dumb charges. 1978 Oxford Diocesan Mag. Dec. 17/1 Then there was the two days' retreat... At no time..was the sense of fellowship more apparent than during the silence. 

50

    f. Proverbial phr. silence is golden. silence gives consent: see CONSENT n. 1c.    1834 CARLYLE Sart. Res. III. iii, in Fraser's Mag. June 668/1 As the Swiss Inscription says: Sprechen ist silbern, Schweigen ist golden (Speech is silvern, Silence is golden). 1865 W. WHITE Eastern England II. ix. 129 Silence is golden, says the proverb. We apprehended the full significance thereof when far away from busy thoroughfares. 1935 M. V. HUGHES Vivians vii. 138 ‘Did you tell him about that?’ ‘No, and I'm wondering whether I ought to?’ ‘I shouldn't if I were you. Silence is golden.’ 1980 J. O'NEILL Spy Game xxv. 239 ‘I'll tell you the rest..on the way back.’ He sealed her lips with a finger. ‘Meanwhile, silence is golden.’ 

    2. a. The state or condition when nothing is audible; absence of all sound or noise; complete quietness or stillness; noiselessness. Sometimes personified. Also const. of (the night, etc.).    1382 WYCLIF Isaiah viii. 6 The watris of Siloe, that gon with cilence. 1398 TREVISA Barth. De P.R. X. ii. (1495) 27b, Derknesse is seen yf noo thynge is seen, & scylence is knowen yf noo thynge is herde. 1500-20 DUNBAR Poems xxxv. 1 Lucina schynnyng in silence of the nicht. 1590 SPENSER F.Q. III. i. 59 Whenas all the world in silence deepe Yshrowded was. 1602 SHAKES. Ham. II. ii. 506 But as we often see against some storme, A silence in the Heauens. c1630 MILTON Circumcision 5 Through the soft silence of the list'ning night. 1738 WESLEY Hymns, ‘Regent of all the Worlds above’ iii, Fair Queen of Silence, Silver Moon. 1784 COWPER Task VI. 84 Stillness, accompanied with sounds so soft, Charms more than silence. 1832 MACAULAY Armada 49 Then bugle's note and cannon's roar the deathlike silence broke. 1850 TENNYSON In Mem. xix, There twice a day the Severn fills;..And makes a silence in the hills. 1878 BROWNING La Saisiaz 25 Can I..sharpen ear to recognize Sound o'er league and league of silence? 

    b. Used allusively to denote the state beyond this life. Chiefly in pl. and with initial capital.    1803-6 WORDSW. Ode Intimat. Immortality ix, Power to make Our noisy years seem moments in the being Of the eternal Silence. 1851 CARLYLE Sterling I. i, To return silently, with his small, sorely foiled bit of work, to the Supreme Silences. 1908 E. MILLER Martyrs of the Moors 55 In fear and darkness his soul floated out to the great Silence. 

    c. Tower of Silence, one of a number of small towers upon the summit of which the Parsees place their dead (see quot. 1865).  

51

  1865 Chambers's Encycl. VII. 300 Their dead are not buried, but exposed on an iron grating in the Dokhma, or tower of Silence, to the fowls of the air. c1880 GRANT Hist. India I. lxix. 359/1 The exposure of their dead in the Towers of Silence, to be eaten by the birds. 

    d. Phr. the rest is silence and varr., in allusion to the last words of the dying Hamlet (SHAKES. Ham. V. ii. 368).    1910 GALSWORTHY Justice II. 49 Once this cheque was altered and presented, the work of four minutes four mad minutes the rest has been silence. 1939 A. HUXLEY After Many a Summer II. i. 187 If only the rest were silence!.. What joy if the rest of Wordsworth had been silence, the rest of Coleridge, the rest of Shelley! 1982 Daily Tel. 2 June 16/4 In most of the countries involved the eternal tug-of-war between Government and news media has long since ended. The curtain has fallen. The rest is silence. 

    e. A period of silence observed in memory of the dead, esp. the two minutes' silence kept on the anniversary of Armistice Day (11 Nov. 1918) or, since 1946, on Remembrance Sunday.    1919 Times 12 Nov. 15/6 The Great Silence... At 11 o'clock yesterday morning the nation, in response to the King's invitation, paid homage to the Glorious Dead by keeping a two minutes' silence for prayer and remembrance. Ibid. 16/1 On the Stock Exchange, after the silence, a gong was sounded. 1926 A. TOPHAM Chron. Prussian Court xx. 245 We discussed among other things the Titanic disaster [1912] which had recently happened, and I remember referring to ‘the silence’ of two minutes by which the Canadian railways and churches had honoured the memory of the Canadians who had perished. 1929 B.B.C. Year-bk. 1930 78 Broadcasting the Silence November 11th, 1928. 1972 ‘E. LATHEN’ Murder without Icing (1973) xxii. 188 The game was preceded by a two-minute silence in memory of Billy Sicagusa. 1982 D. PHILLIPS Coconut Kiss vi. 52 You march once round the playground and salute the flag... Then you go in for the two minutes' silence. 

    3. a. Omission of mention, remark, or notice in narration. Chiefly in phrases to pass with, pass over in, silence.    1513 BRADSHAW St. Werburge Prol. 114 It were no reason her name be had in scylence, But to the people her name be magnyfyed. 1585 T. WASHINGTON tr. Nicholay's Voy. IV. xxxiii. 156 Ordinances, which I passe with silence. 1600 J. PORY tr. Leo's Africa III. 131, I would much rather haue smothered such matters in silence. 1667 MILTON P.L. VI. 385 Eternal silence be thir doome. 1711 ADDISON Spect. No. 1 3 As for the rest of

52

my Infancy, there being nothing in it remarkable, I shall pass it over in Silence. 

    b. Neglect or omission to write (about something); failure to communicate or reply.    1617 MORYSON Itin. II. 206 By Don Jeans silence from Spaine, this overture..tooke no effect as long as the Queene lived. 1698 FROGER Voy. Aj, The silence of all those who made the voyage with me, constrained me to expose it to publick view. 1771 Junius Lett. liv. (1788) 292, I under~stand that the public are not satisfied with my silence; that an answer is expected of me. 1790 PALEY Horæ Paul. II. i. 9 The silence of the historian..concerning any contribution, might lead us to look out for some different journey. 

     4. A small hammer used to command silence or order. Obs. 1    1556 in Jupp Acc. Carpenters' Comp. (1887) 139 He helde not his peess before the master hade knockyd with the sylence iij tymes. 

    5. Mus. A rest.    1752 tr. Rameau's Treat. Music 171 This Silence or Rest can be made but upon a Concord or consonant Note. 1856 MRS. BROWNING Aur. Leigh v. 342 The soul..With all its grand orchestral silences To keep the pauses of its rhythmic sounds. 

    6. Distill. Want of flavour in distilled spirit. (Cf. SILENT a. 5c.)    1879 Spon's Encycl. Manuf. I. 229 Owing to its ‘silence’, there is no possibility of detecting afterwards from what source it has been obtained. 

    7. attrib. and Comb., as silence time, and in recent use silence-box, command, room, rule, etc.; silence-loving adj.; silence cabinet, (a) = silence-box; (b) (see quot. 1929).    1889 Telephone I. 471/1 The public is also admitted to a silence-box at the Nottingham Post Office. 1894 Daily News 28 Mar. 3/1 These particulars are telephoned into a silence-box at the Central Savings Bank.

53

1893 PREECE & STUBBS Man. Teleph. 227 At most telephone exchanges a ‘silence cabinet’ is provided in the public office. 1929 B.B.C. Year-bk. 1930 309 In Savoy Hill there are nine studios, six of which are equipped with silence cabinets (these are small rooms adjacent to the studios from which the announcer can speak before switching over to the studio itself).

1855 F. W. FABER Growth in Holiness ix. 147 It wrung a cry even from the silence-loving Heart of our ever-blessed Saviour. 1912 W. OWEN Let. 23 June (1967) 142 The firm Superintendent of their Sunday School, the silence-loving, and the melancholy-voiced, on that day capered about the lawn among them. 1958 S. HYLAND Who goes Hang? xviii. 77 They were in the Silence Room of the Library, a room in which conversation..is..a tabu. 1959 T. S. ELIOT Elder Statesman II. 47 And remember, when you want to be very quiet There's the Silence Room. With a television set.

1894 Daily News 14 May 5/1 He promised the modification and virtually the abolition of the silence rule.

1387 TREVISA Higden (Rolls) V. 19 Speke wolde he nevere, as it is i-write in e questiouns at he wroot in his scilence tyme. c1450 in Aungier Syon (1840) 268 Eche suster..shalle answer thus a ene in lyke voyce thof it be sylence tyme.

54

Appendix 2 – Copy of Kalman & Rafaeli, 2005

55