Sik hung ng; loong, cynthia s. f.; an ping he; liu, james h.; weatherall, ann
Sik Hung Ng; Loong, Cynthia S. F.; An Ping He; Liu, James H.; Weatherall, Ann
Journal of Language & Social Psychology, Mar2000, Vol. 19 Issue 1, p26, 20p
COMMUNICATION CORRELATES OF INDIVIDUALISM AND
Talk Directed at One or More Addressees in Family Conversations
Conversations in 12 European and 12 Chinese families in New Zealand were recorded and a sample of 15,988
turns of talk was coded under single-and multiaddressee turns. As predicted, the two types of turns varied with
in both inter- and subcultural comparisons. Single- addressee turns were
more common in European (individualistic) than in Chinese (collectivistic)families, and in high (individualistic)
than in low (collectivistic) acculturated Chinese families; the reverse was true for multiaddressee turns. The
predicted relationship was confirmed by an individual- level comparison: Members of Chinese families who
were highly acculturated used more single- and fewer multiaddressee turns when compar ed with less highly
acculturated members. This relationship was unaffected by language (English or Chinese) or age of family
members. In a separate analysis of the effect of turn types on turn taking, it was found that multiaddressee turns
were associated with more false starts (two or more members started talking simultaneously). Implications of
the results for cross-cultural and communication research were discussed.
constitute one of four basic dimensions for describing and comparing cultures
(Hofstede, 1980). Their importance in cross-cultural research, more so than the other dimensions, has been
attested to by four major reviews in the last decade alone (Bond & Smith, 1996; Kagitcibasi & Berry, 1989;
Singelis, Triandis, Bhawuk, & Gelfand, 1995; Triandis, 1990). Most of the studies on cultural individualism
have been published in social (cross-cultural) psychology. In a separate development
that has occurred mainly outside social psychology, communication has emerged as an important area for cross-
cultural comparison (e.g., Chang, 1997; Gudykunst, Ting-Toomey, & Nishida, 1996; Singelis & Brown, 1995).
Our present research is primarily concerned with family conversations and how ways of communicating may
differ between cultures; in addition, we also attempt to explore links between communication and cultural individualism
. Below we first review some of the major approaches to the culture-communication
nexus and then propose a novel link based on the concept of addressee orientation.
One major existing approach is to start from the social relational demands of individualism
and proceed to theorize how those demands are reflected or reproduced in interpersonal communication.
Reviews of relevant studies by Clyne (1994), Gudykunst and Matsumoto (1996), Scollon and Scollon (1995),
and Triandis (1994) have painted the following picture. Individualistic cultures emphasize the self over the
group and personal independence over social interdependence. Those cultural values encourage self-expression
and speaking one's mind freely. While speaking, the speaker draws attention to the self and follows Grice's
(1975) cooperative principle: "Make your conversational contribution such as required, at the stage in which it
occurs, by the accepted purpose or direction of the talk-exchange in which you are engaged" (p. 45). By
contrast, in collectivistic cultures, people grow up wedded more strongly to their groups and learn to value
interdependence more than personal independence. As a result, the anticipated social implications of what one
says will be given equal, if not more, weight than following the Gricean cooperative principle of speaking one's
own mind, no matter what. This leads to self-censoring, and even compromised talk, for the sake of maintaining
social harmony, respecting the existing status hierarchy, and so forth. Essentially the same argument has been
advanced to explain cultural differences in interruptions (Murata, 1994) and why positive and negative
politeness, as defined by Brown and Levinson (1987), are used differently across cultures (e.g., Gu, 1990;
Another approach is exemplified by Hall's (1976) pioneering works on low- and high-context communication. In low-context communication, the wider context surrounding a message plays a relatively minor role compared to the message itself: Information encoded mainly in the message is often direct and explicit in its meaning. The burden of interpersonal communication is then on the speaker to construct and deliver messages that conform to Grice's (1975) cooperative principle. Speakers are expected to tell the truth, concisely but fully, to the point and unambiguously, rather than to skirt around or to compromise it for such purposes as maintaining group harmony. In high-context communication on the other hand, relatively little is transmitted in the explicit message itself, whereas most of the information is embedded in the context or internalized by listeners who are expected not only to listen but also to infer the speaker's intention from what is not explicitly said.
Individuals from the same culture may use low - or high-context communications, depending on the situation
and the person with whom they are talking. For example, formal, especially legal, situations would favor the use
of low-context communication and also when speakers are communicating with strangers with whom they have
little shared common ground. In addition to the formality of a situation and interpersonal relations, culture also
plays a part in influencing the relative use of those two styles of communicating. As Hall (1976) has
perceptively noted, low-context communication is more strongly associated with cultural individualism
high-context communication with cultural collectivism
. Later studies that lend support to Hall's insight have
been summarized by Gudykunst and Matsumoto (1996).
Hall's (1976) approach, which is primarily anchored in communication but also allows for exploration into
, will be adopted for the present article. Our point of departure from Hall is
that instead of looking at communication in terms of direct and indirect meanings constructed from the interplay
between message and context, we will focus on turn-taking behavior in group conversations. Turn taking, no
less than meaning making, is fundamental to interpersonal communication (Clark, 1985); however, unlike
meaning making, its implications for individualism
remain to be explored. Essentially, we
propose that there are two principal addressee orientations in turn taking that are interesting in themselves and
may also provide a novel way of comparing cultural individualism
Bales is well-known to social psychologists for his works on communication that deal with such questions as "Who talks to whom, how often" in group interactions (Bales, 1950). In social psychology, those questions are pertinent to leadership role differentiation (Hollander, 1985), group response to deviants (Schachter, 1951), group dynamics (Shaw, 1981), and social influence in both group (Ng, 1996) and intergroup (Reid & Ng, 1999) settings. They presuppose that speakers are addressee oriented in their talk--that they are talking to another person or persons, rather than talking past each other or talking to themselves. That presupposition is often taken for granted in social-psychological research and sometimes even reified in controlled laboratory settings in which research participants are instructed to address their written messages to each other. The important issue here is not so much that the assumption of addressee-oriented talk does not apply to verbal interaction in real life--it clearly does --but, rather, the subtlety of addressee or ientation has been overlooked.
From a communication perspective, broadly defined here to include conversation analysis (Sacks, Schegloff, &
Jefferson, 1974; see also Ng & Bradac, 1993), addressee-oriented turns have the potential of selecting or
engaging particular listener(s) in the group as next speaker. Just as significantly, they have the potential of
discouraging or excluding nonselected others from speaking. Thus, depending on the scope of their affiliative or
engaging focus, turns with different addressee orientations vary in their social inclusiveness or exclusiveness.
Single-addressee turns have a relatively exclusive, individualistic focus. By contrast, multiaddressee turns that
are directed to two or more persons have a more inclusive, collectivistic focus. It is this exclusiveness-
inclusiveness dimension that has been overlooked in social-psychological studies of group interaction and yet
may provide a link with cultural individualism
Specifically, we predict that turns directed to a single addressee would be higher in families from an
individualistic than from a collectivistic culture and, conversely, for turns that are directed to more than one
addressee. The proposed effects of culture will be examined by both inter - and subcultural comparisons. First,
we test for differences in single- and multiaddressee turns between Chinese and European families in New
Zealand. It is assumed that these two ethnic groups represent cultural collectivism
respectively. This assum ption is based on Hofstede's (1980) survey, which placed New Zealand Europeans near
the top of the individualism
scale and Hong Kong and Taiwanese Chinese near the bottom of that scale. A
similar comparison was found in a more recent study showing New Zealand Europeans to be more
individualistic than Chinese in Hong Kong and especially in mainland China (Ng & Zhu, 1999; see also
Scherer, 1997). New Zealand Chinese were not represented in those studies, although most of them shared the
same cultural background as Chinese in Hong Kong, China, and Taiwan. Following on from intercultural
(interethnic) comparisons, the effect of culture is tested further at a subcultural level by comparing low
acculturated (recent immigrant) Chinese families with high acculturate d Chinese families who have lived in New Zealand for two or more generations. Single-addressee turns were expected to be higher in European than in Chinese families and in acculturated Chinese than in recent immigrant Chinese families, whereas the opposite results were expected for multiaddressee turns.
As the study progresses, comparisons at the level of individuals will also be developed and reported in the present article.
OVERVIEW OF STUDY
The data were sourced from a wider study that included 20 European and 18 Chinese families with three generations now living in the Wellington region (Ng, Liu, Weatherall, & Loong, 1998). Conversations in all of the 20 European families involved two members of each of the three generations of grandparents, parents, and grandchildren. Only 12 of the Chinese families satisfied that condition, the remaining families lacking either one grandparent or one grandchild. For ease of intercultural comparisons, family size and members' ages were matched as closely as possible. The final sample for the present study therefore consisted of 12 six-member families of each ethnic group, with the grandparents all older than 65 years, most of whom being in their late 60s or 70s. The middle-aged parents were between 35 and 50 years, whereas the youngest generation (grandchildren) were in their teens or early 20s, although a few were younger than 10 years. Typically, a male and a female from each generation took part in the conversation, but in 9 families, gender was uneven for the youngest generation, with either all being female (2 Chinese and 2 European families) or all male (2 Chinese and 3 European families).
All European participants had been born in New Zealand, except for one grandparent who had been born in Yugoslavia. Most of the Chinese families had migrated from south China or Hong Kong, although one was from Malaysia and another from Taiwan. Length of residence in New Zealand varied from 2 years to fourth-generation New Zealanders. English and the Chinese dialects of Cantonese, Seyip, and Mandarin were all represented in the Chinese sample. In 10 families, all members were of Chinese descent and in the remaining 2 families, the middle-aged father was a European.
The 12 Chinese families were divided into high- and low-acculturation groups on the basis of place of birth and length of residence in New Zealand. There was a clear break between families with four or more members born in New Zealand and the remainder, which had only two or fewer members born in New Zealand. On that basis, the former set was designated high acculturation and the latter low acculturation. As it turned out, the two families that had a European father were classified in the high- acculturation group. There were 6 families in each acculturation group.
RECORDING FAMILY CONVERSATIONS
Recordings of family conversations were made during family gatherings involving all three generations, in the privacy of their homes. Two recording sessions were made, in one of which the six family members were asked to talk about any topics they liked. In the other session, a list of topics was provided that they could use if they so wished. Each session lasted at least 30 minutes, and for most families the total recording time was about 120 minutes. Those two sessions were recorded during the same day or on separate days.
Each family member wore a wireless radio microphone for transmitting their voice to a multitrack audio recorder located out of sight in another part of the house. This recording arrangement freed family members to move around while being recorded. Separate channels on the tape were dedicated to different family members to facilitate identification of speakers when their conversations were later transcribed. Assurance was given to participants prior to the start of the recording that they could switch off their microphones at any time and that if there was anything that they did not want to remain on the tape, they could delete it before the tape was handed over to the researchers. One or two researchers stayed in the house to operate the recording equipment and, when needed, to answer questions.
TRANSCRIBING AND CODING FAMILY CONVERSATIONS
The taped conversations were transcribed for their content, the identity of speakers, and their speech delivery, using transcription conventions adapted from Button and Lee (1987). As some speakers could use either or both
of Chinese and English, their actual language was transcribed verbatim by means of a bilingual word-processing software. To check the accuracy of each transcription, all transcripts were proofread against the tape recordings.
For each family, a total of 30 minutes of their conversation was selected for coding and analysis. Selection was based on continuous conversation segments, each about 3 minutes long. Six to 12 segments were selected at approximately even intervals to cover the full transcript. Selection and coding (see below) were carried out for the main study by assistants blind to the hypotheses expressed in the present study.
Prior to the coding, two coders listened to the whole audio recording of each family conversation to become thoroughly familiar with the flow of the conversation and with the transcript. Only then did the coding begin. The continuous stretch of an utterance spoken by a family member up to the point of speaker switch was coded as a single turn if it had at least one complete meaning. Minimal responses such as "ah ah," "yeah," and "mm" were not counted as turns. Each turn might occur alone or in overlap with the utterance of another speaker(s). Such overlapping utterances were counted as separate turns. If the overlap resulted in a pause of longer than I second, the continuation of the postpause utterance was counted as a second turn; otherwise, the whole utterance was coded as a single turn.
Each turn of talk was coded as "addressee oriented" if it satisfied either one or both of two criteria. The first was personal addressing, when a turn that addressed one or more particular persons by name or another identity marker was counted as addressee oriented (Sacks et al., 1974). The second criterion was topical coherence or relevance, when a speaker addressed the same topic raised or developed by another speaker (Schiffrin, 1994). The immediately previous speaker was usually, but not necessarily, the topic -raiser or -holder. Sometimes, however, it was necessary to track further back in the conversation to identify the most relevant prior speaker's topic. Note that this linking exercise was flexible enough to trace the topic to more than one prior speaker. For example, within a turn, a speaker might talk about a topic that had been talked about by another speaker and then move to another topic that had been raised by a different speaker. Each addressee- oriented turn was further classified as either a single - or a multiaddressee turn. Where neither personal addressing nor topic coherence was evident in a turn, the turn was not considered addressee oriented and was coded as "unclear." Two coders who had been trained for the task and were thoroughly familiar with these family conversations carried out the coding. Although they knew the conversants' ethnic origin and generation, they were blind to the hypotheses at the time of coding. They also cross-coded 135 turns to check intercoder reliability in the use of the three coding categories. Cohen's Kappa = .76, which indicated a satisfactory level of agreement.
PART 1: INTER- AND SUBCULTURAL COMPARISONS
The data set comprised 15,988 turns from the 24 families. The numbers of turns per family varied from 570 to 843 (median = 652) in the European sample and from 398 to 853 (median = 625) in the Chinese sample. For number of turns, t tests showed no significant differences either between ethnic samples or between high- and low-acculturated Chinese families. (In all the analyses reported here, the significance level was set at p < .05 unless otherwise stated.) But in view of the large variations across families, the numbers of single - and multiaddressee turns within each family were expressed as percentages of the family total to form the basis of data analyses. Note that these two percentages did not sum to 100 because each family's total turns also contained turns that were coded "unclear."
Prior to the main data analyses, the data were screened for outliers and normality of distribution. This was deemed necessary because of the small number of families. There were no obvious outliers, and both kurtosis and skewness tests indicated no serious departure from normality (all tests resulted in absolute values of less than 1). Further inspection of the data showed a negative correlation between single- and multiaddressee turns (Pearson's r = -.87, n = 24), suggesting that a certain amount of redundancy is inevitable if both types of turns are to be analyzed. However, instead of using only one turn type for the analyses, both types are used to allow for a fuller evaluation of the hypotheses.
To provide the fullest possible context for the inter- and subcultural comparisons, the three groups (European families, high-acculturated Chinese families, and low-acculturated Chinese families) were treated jointly as a between-subjects variable in a one-way analysis of variance (ANOVA) design. The ANOVA showed a significant main effect for groups, F(2, 21) = 5.83, MS[sub error] = 66.28, in the predicted direction (upper part of Table 1). For the intercultural comparison, first the mean percentage of turns among Chinese families as a whole was calculated from the high- and low-acculturated figures. This Chinese mean (M = 57.3) was significantly lower than the European mean (M = 66.4), F(1, 21) = 7.40. For the subcultural comparison
between high- and low -acculturated Chinese families, the result confirmed the prediction of a significantly higher mean among high- than among low-acculturated families (Ms = 62.2 versus 52.5), F(1, 21) = 4.25,p = .052.
For multiaddressee turns, a one-way ANOVA showed a significant main effect for groups, F(2, 21) = 5.55, MS[sub error] = 54.50. The means, as shown in Table I (lower portion), were in the predicted direction. Unlike the case of single-addressee turns, here the intercultural comparison was not significant, F(1, 21) = 3.83, p > .05. The subcultural comparison remained significant, F(1, 21) = 7.15.
The results of both the inter- and subcultural comparisons converge in showing a positive relationship between
single-addressee turns and cultural individualism
, and between multiaddressee turns and cultural collectivism
At the intercultural level, single- addressee turns were proportionally more common, and multiaddressee turns
less common, in European than in Chinese families. Among Chinese families, single-addressee turns increased,
and multiaddressee turns decreased, with family acculturation. Thus, by treating the family as a conversational
system, the results show that the pattern of single- and multiaddressee turns corresponds systematically to the
family's cultural background and, among Chinese, to its level of acculturation. Although the number of families
is small, the confluence of results gives credence to the hypothesized effect of cultural individualism
PART 2: INDIVIDUAL-LEVEL COMPARISONS
The results from subcultural comparisons demonstrate the effect of acculturation at the level of families considered as a whole. To test if that effect also operated at the level of individual family members, a comparison was made by using generation as a proxy variable for individual acculturation among Chinese. The assumption was that the youngest generation, who were mostly born and brought up in New Zealand, would be the most acculturated, whereas the oldest generation would be the least acculturated for the opposite reason. The middle-aged generation would be intermediate in acculturation. One would therefore expect single-addressee turns to increase, whereas multiaddressee turns would decrease, from grandparents to parents and to grandchildren.
For each family member, the numbers of single- and multiaddressee turns were expressed as percentages of that member's total turns, to control for variation in total turns across members.
Kurtosis tests of the percentages of single- and multiaddressee turns showed unacceptably high values of 1.6 and 2.7, respectively. A screening of the data set showed one outlier --a middle-aged mother in the low-acculturated category who had more than 99% multiaddressee turns, well above all the rest. After this person was excluded from the data set, the kurtosis values were reduced to -.05 and -.41, respectively. (Skewness was less than I in absolute terms regardless of whether the outlier was included or excluded.) That outlier was therefore excluded from the correlational analysis and ANOVAs below.
Single- and mult iaddressee turns were negatively correlated (Pearson's r = -.93, n = 71). Following the same rationale as in Part 1, ANOVAs were carried out on both types of turns. To provide the fullest possible context for evaluating the hypothesized generation effect, gender and high- versus low-acculturated families were added to form a three-way ANOVA design. In both types of turns, the results showed two significant main effects and no interaction effects.(n1)
Single-addressee turns. The significant main effects wer e family acculturation, F(1, 59) = 11.24, and generation, F(2, 59) = 12.87, MS[sub error] = 137.19. The former (see means in the upper left part of Table 2) replicated the subcultural comparison results found previously at the family level. The generation effect supported the hypothesis concerning the effect of acculturation at the individual level because single- addressee turns were the highest among the youngest generation and the lowest among the oldest generation (upper right part of Table 2). Comparisons of cell means by a Newman-Keuls test showed that the means of the youngest and the middle generations did not differ significantly from each other, but both were significantly higher than the mean of the oldest generation.
Multiaddressee turns. As was the case for single-addressee turns, multiaddressee turns showed significant effects for family acculturation and generation. The former effect, F(1, 59) = 24.52, shown in Table 2 (lower left), replicated the corresponding subcultural comparison reported earlier under Part 1. The main effect of generation, F(2, 59) = 18.07, MS[sub error] = 102.31, supported the hypothesis of individual acculturation. Newman-Keuls tests of the three means (Table 2, lower right) showed that the mean of the oldest generation was significantly higher than the mean of the middle generation, which, in turn, was significantly higher than the mean of the youngest generation.
The overall results replicate the subcultural finding of the effects of acculturation on turn types among Chinese. Single-addressee turns increase from the less highly acculturated members (grandparents) to the more highly acculturated members (parents and grandchildren) of the family. Conversely, multiaddressee turns decrease from grandparents to parents and from parents to grandchildren. However, a note of caution should be introduced at this point because it is possible that the effect of generation is not due to acculturation but to age per se. To test this, the European data were analyzed with reference to generation and gender in a two-way ANOVA design. If the effect of acculturation that has been hypothesized to correlate with generation in Chinese families was merely an age effect, the European results should also show a significant generation effect similar to that for the Chinese. But no such effect was found for either type of turns. Across European generations, the percentage of single- addressee turns varied only slightly from 64.0 (parents) to 68.2 (grandchildren), whereas multiaddressee turns varied from 29.0 (grandchildren) to 32.3 (parents). These negative results (in the European sample) make it unlikely that the significant effect of generation in the Chinese sample is an artifact of age.
The data analytic scheme for the individual- level comparisons above was based on the assumption that all six members of the same family were independent units. That assumption is problematic because the speech of an individual depended on the speech of other family members, and for this reason the F ratios and p levels might have been biased. To check the extent of bias, a different analysis was conducted in which the family members were treated as repeated measures (generation by "gender") nested within families. The repeated measures were combined with ethnicity to form a 3 x 2 x 2 mixed- ANOVA design. (In the few families with two male or two female children, "gender" for children was randomly assigned.) For both single- and multiaddressee turns, the ANOVAs showed the same three significant effects: ethnicity, generation, and Ethnicity x Generation interaction. The interaction effect was F(2, 42) = 6.04 (MS[sub error] = 112.53) for single-address turns and F(2, 42) = 6.58 (MS[sub error] = 85.53) for multiaddressee turns. Using the new error terms and ns (12 European and 11 Chinese families, excluding 1 Chinese family with the outlier referred to earlier), planned comparisons confirmed the previous results of a generation effect among Chinese but not European families. Specifically, in Chinese families, grandparents' single- addressee mean was significantly lower than parents' or children's mean, Fs(1, 42) > 5.92, and parents' mean was lower than children's but not significantly so. For multiaddressee turns, grandparents' mean was significantly higher than parents' mean, F(1, 42) = 5.32, which, in turn, was significantly higher than children's mean, F(1, 42) = 4.54. In European families, on the other hand, none of the comparisons were significant.
Although the results presented above seem robust, they might have been confounded by the different languages used in family conversations. In Chinese families, English was used more extensively by the more highly acculturated families, whereas Chinese was more often used by the less well acculturated families . Because of those systematic linguistic variations, language could have confounded acculturation in producing the results. To explore this issue, further analyses were carried out to reexamine the data for the possible effects of language.
European families were excluded because they all spoke English. For each member of the Chinese families (excluding the outlier referred to earlier), the number of English turns was counted. An English turn was defined initially as a turn that was spoken wholly in English. This led to a skewed distribution with several family members having zero English turns. To reduce skewness, the definition was relaxed to include turns that were spoken partly in English. Each member's English turns were expressed as a percentage of the member's total number of turns for analysis. The percentage data were transformed by natural log to normalize the distribution (kurtosis = .07, skewness = -.86). Results based on the transformed data, to be reported below, matched closely those based on the raw data with identical significant findings.
Preliminary analyses were carried out to correlate English turns with generation and acculturation. Generation was coded either 1 (grandparents), 2 (parents), or 3 (grandchildren) for analysis by Pearson correlation, the result of which was near zero (Pearson's r = .09, n = 71). Acculturation was a dichotomous variable (low acculturation = 1, high acculturation = 2), and for this reason point-biserial correlation was used. The result
showed that acculturation was significantly correlated with English turns (point-biserial correlation = .91, n = 71).
Analysis of covariance. Two sets of main analyses were conducted. First, the percentage of English turns was used as a covariate to reanalyze the effects of gen eration, gender, and family acculturation. For single-addressee turns, an analysis of covariance showed a significant main effect for generation that was uncomplicated by either gender or acculturation, F(2, 58) = 12.65. Comparisons of the adjusted means by a Newman-Keuls test replicated the findings that had been summarized in Table 2. A similar straightforward main effect of generation was found for multiaddressee turns, F(2, 58) = 17.67. Newman-Keuls tests on the adjusted means again replicated the corresponding tests in Table 2.
Regression analyses. In the second set of analyses, the aims were to test whether English usage affected the types of turns independently of generation and to compare the respective effects of English turns and generation. Hierarchical regression analyses were carried out by entering English turns first and then generation (see Table 3). For single-addressee turns, 9% of the variance (R[sup 2]) was accounted for by English turns alone (Step 1), and another 21% was added when generation was entered (Step 2). The change of R[sup 2] for Step 2 was significant. The results in Step 2 showed that generation had a beta weight (Beta) higher than that of English turns. A similar pattern of results was obtained for multiaddressee turns (see Table 3).
Thus, when English usage is controlled for in the analysis of covariance, the results confirm the main effects of generation. The regression results show that in Chinese families the use of English affects both single- and multiaddressee turns and yet does not make the effect of generation redundant. Because of the limited scope of this study, it is not possible to tell if the English effect is due to language per se or an experience of acculturation that covaries with the acquisition of English.
The zero correlation between generation and English reported earlier is intriguing and suggests opposing patterns of correlation between subgroups within the Chinese sample. Further analysis showed that English was correlated with younger generations positively in low-acculturated families (Pearson's r = .38, n = 35,p < .05) but negatively in high-acculturated families (Pearson's r = -.31, n = 36,p = .06).
PART 3: TURN TYPES AND "FALSE STARTS"
The analyses so far have concerned single- and multiaddressee turns as a "dependent variable," to be explained
by cultural individualism
and acculturation. We would want to extend the research by exploring a
theoretical implication of the types of turns for group conversations. As a preliminary attempt in this direction,
we start from the classic model of conversational turn taking proposed by Sacks et al. (1974). This model makes
the strong claim that "overwhelmingly, one party speaks at a time." Overlapping speech in which two or more
speakers start speaking at the same time (false start) may occur but is supposed to be brief and will be quickly
repaired to restore the convention that only one party speaks at a time. Within this model, false starts in one-to-
one conversations are attributed mainly to mistaken interpretations of the speaker's intention to pause for
continuation or to pause for yielding the turn, often aggravated by nonverbal signals (Duncan & Fiske, 1977).
Similar reasons may also apply to false starts in group conversations, but here we think multiaddressee turns
have a role to play too. The collectivistic focus of multiaddressee turns invites speakership from more than one
person. This does not necessarily lead to a false start in the next immediate turn, but in family conversations,
with a greater proportion of multiaddressee turns, false starts are more likely to occur overall. If that reasoning
is valid, one would expect to find first, a positive correlation between multiaddressee turns and false starts, and
second, a greater frequency of false starts in Chinese than in European families and in low - than in high-
acculturated Chinese families.
A "false start" can be operationalized as a segment of conversation where two or more parties start talking together after a prior speak er has finished a turn of talk. It is a type of overlapping talk characterized by its position at the beginning of and not during a turn. As such, it differs from interjections, minimal responses, interruptions, and other overlapping talk that occurs well after a speaker has commenced a turn (see Roger, Bull, & Smith, 1988). To qualify as a false start in the present study, the speakers involved must start talking at the same time or within one syllable of another. This rule was adopted when the family conversations were transcribed, allowing us to calculate for each family the total number of false-start turns from the transcript.
The number of false starts in a family was expressed as a percentage of the total family turns. To test whether multiaddressee turns increased false starts in family conversations, the percentage of multiaddressee turns was correlated with the percentage of false starts. As predicted, the correlation was positive and significant (Pearson's r = .35, p <.05, n = 24). Consistent with this finding, a significant negative correlation was found between false start and single-addressee turns (Pearson's r = -.36,p < .05, n = 24). An intercultural comparison of false starts showed a significantly higher mean in Chinese than in European families (M[sub Chinese] = 10.9, M[sub European] = 6.2), F(1, 22) = 10.31. A subcultural comparison of low- and high- acculturated Chinese families showed no significant difference.
Our concept of addressee orientation shares with other similar concepts such as other orientation and receiver centeredness the notion that speakers attend or attune to other group members (Ambady, Koo, Lee, & Rosenthal, 1996; Yum, 1988). It differs from them in making a psychologically important distinction between a single- and multiaddressee focus. The point we are making is that it is not so much that some cultures are addressee oriented and others are not but that they differ in terms of how frequently a more or a less inclusive focus will be used in group conversations. Multiaddressee turns, being more inclusive, are more often used in collectivistic than individualistic cultures and vice versa for single-addressee turns. This central thesis is supported by all three comparisons undertaken in the pres ent study. Single-addressee turns are more common in New Zealand European families than in New Zealand Chinese families, in high- than in low-acculturated families in the Chinese community, and for high- than for low-acculturated members of Chinese families. Multiaddressee turns show the reverse pattern of comparisons. The comparisons are pitched at different levels, yet the results are remarkably convergent.
It is important to stress the research strategy of multilevel comparisons, not only because of the coherent pattern
of results it has produced but also in recognition of the nonexperimental nature of the present research. As
Messick (1988) has pointed out, cross-cultural research seldom manipulates culture as an independent variable
or assigns research participants randomly to different cultural conditions. For that reason, "it is nearly
impossible to ensure that one and only one factor differs between cultures" for valid comparisons to be made (p.
45). The criticism is fair and also applies to the pres ent research. Admittedly, we are in no position to rule out
the possibility that variables other than cultural individualism
may have been responsible for the
results. This uncertainty is greatest at the intercultural level of comparison--New Zealand Europeans and
Chinese differ in so many ways that any number of their differences may have caused the between-group
differences in turn types. However, as we develop more specific comparisons at the subcultural and individual
levels, the sources of extraneous variables are progressively reduced, and the validity of cultural individualism
is correspondingly enhanced. The fact that the results are not confounded by age
(generation) per se or language makes the convergent findings all the more compelling. In short, the painstaking
coding of conversational units has made a novel contribution to research that attempts to demonstrate the
behavioral correlates of cultural individualism
without having to rely on self -reported or attitudinal
Apart from providing a behavioral measure for cultural comparison, the concept of addressee orientation also has implications for false starts and, more generally, turn taking. Previous research has shown that nonverbal signals (Duncan & Fiske, 1977) and familiarity among conversants (Dunne & Ng, 1994) affect the frequency of false starts. The present study shows that addressee orientation may also have a role to play, as evidenced by the finding that multiaddressee turns increase and single-addressee turns decrease false starts. The results suggest that Sacks et al.'s (1974) one-party-speaks- at-a-time model of conversation, which acknowledges false starts but maintains that they are relatively rare, may not be applicable to group interactions in which multiaddressee turns are widely used. More work is needed to clarify how multiaddressee turns may encourage false starts and affect the repair of false starts.
The implication of addressee orientation for turn taking goes beyond false starts. The varying degrees of social inclusiveness associated with single- and multiaddressee turns may generate different dynamics for group interaction. The case in point is the phenomenon of "floor state." A floor state refers to a three-turn interactional sequence in which the same two group members, A and B, alternate speaking: A arrow right B arrow right A. There is evidence to show that the floor state is the most common pattern of interaction in four-person groups (Parker, 1988). Even in six -person groups, where as many as four other persons can potentially prevent the development of a floor state by displacing A at turn 3, the floor state still occurs frequently and has more than a 50% chance of renewing itself to a fourth turn: A arrow right B arrow right A arrow right B (Stasser & Taylor, 1991).
In a floor state, the three turns can in principle be either single- or multiaddressee turns, or a combination of both. The interesting thing is that the continuation of a floor state would be problematic if the fourth and subsequent turns are multiaddressee rather than single-addressee turns. Excerpt A illustrates the problem. The conversational context surrounding the excerpt was set earlier by Mother, who raised the question of how to spend a sum of $300. Various suggestions were made by members of the family. At one point in the conversation, Grandpa called the family's attention to his suggestion with a multiaddressee turn (Turn 1, see below). Grandma (Turn 2) and then Grandson (Turn 3) replied in a single-addressee format. During the next three turns Grandpa and Grandma codeveloped a floor state between them, beginning with a multiaddressee turn (Turn 4) that was followed by two single-addressee turns. Grandma then continued, not with a single-addressee turn but an open, multiaddressee turn (Turn 7). At this point, Grandson took over from Grandpa, thus stopping the continuation of the Grandma-Grandpa floor state.
Excerpt A 1 Grandpa: hey (1.0) what about mine^ Mother: ((laughter))
2 Grandma: you YOU [you' re outvoted] 3 Grandson: [YOU you had ] your turn 4 Grandpa: i voted [have i ]^ 5 Grandma: [you're out]voted there's[four to two] 6 Grandpa: [o::h i see ](0.5) right ((pause)) yeah arrow right 7 Grandma: so whe (.) when do we go 8 Grandson: mmm^ (.) we'll go to valentines (0.5) couple of weeks
Now contrast Excerpt A with Excerpt B below. The context of the conversation here was Father's visit to a school in Blenheim. At Turn 1, Grandpa began by addressing Father with a question (a typical single-addressee turn). A floor state developed quickly and continued until the end of Turn 8.
Excerpt B 1 Grandpa: what sort of response did you get from the kids^
3 Grandpa: what age (.) would [they be]^ 4 Father: [quit e ] fascinated [I think they] 5 Grandpa: [secondaries] s'pose 6 Father: u::m (1.0) well (0.5) the other day we did one for (0.5) u::m (1.0) and area school in Colvedon (0.5) and so it was (0.5) standard two right up to (0.5) form seven 7 Grandpa: area school I haven't heard that expression before is it^ ((pause)) 8 Father: it is just a school that (0.5) have both primary and secondary school Grandpa: mmm Grandson: mmmmmmmm 9 Grandma: in a in a rural area presumably
The point to note is that the above floor state between Grandpa and Father, which recycled during several turns to the exclusion of other family members, is made up entirely of single-addressee turns.
Excerpts A and B are intended not as proof but merely to illustrate the theme that single- and multiaddressee
turns differ in their social inclusiveness and, for this reason, are useful for comparing cultural individualism
and for shedding light on group interaction. Social inclusiveness, defined herein,
applies to every individual speaking turn and not the aggregate enactment of turns during the course of a
conversation. There is a fine line between the two definitions. A speaker wanting to be inclusive in speaking
may, according to the latter definition, address different group members during separate turns to ensure that
even the shy ones are included in the conversations. Some of his or her turns may be single- and others may be
multiaddressee oriented, but overall the turns are addressed to every and all other group members. To see how
this interesting concept might apply to the family conversations, we randomly selected a transcript from each
cultural group and calculated how many family members were addressed and how many were left out from the
turns of each family member. Without exception, all 12 speakers from the two families included every other
family member in their talk during the course of the conversation. In retrospect, the uniform results are hardly
surprising because the conversations were long enough to allow ample opportunities for group members to talk
to everyone else. A more sensitive measure would be necessary and this awaits further research.
We conclude by revisiting cultural individualism
, a complex concept of considerable value to the
study of social life in its cultural context. In line with Wheeler, Reis, and Bond's (1989) idea of studying individualism
in everyday life, the present research looks at one of the most mundane activities
of everyday life, namely, group conversations. By studying actual conversational behavior rather than relying on
self-reported or attitudinal measures, this research attempts to show how Chinese- and Euro-New Zealanders
manage "being collectivistic" and "being individualistic" in their interaction with family members (see Chang,
1999, for a fuller formulation of the shift in approach). It suggests that ways of communicating are a routine but
an important facet of individualism
, and demonstrates that some of these can be fruitfully
studied by distinguishing between single- and multiaddressee turns and applying them to group interactions in
different cultural settings. A question that remains unexplored in the present research, but is clearly worth
pursuing in the future, is which particular aspects of cultural individualism
are most directly related
to addressee orientation. In the context of New Zealand Chinese, for example, some measure of subjective
acculturation would be useful to track changes in addressee orientation. Finally, addressee orientation appears to
have considerable impact on group interaction generally, and this too points the way for further research.
AUTHORS' NOTE: This research was supported by a grant from the New Zealand Foundation for Research, Science and Technology (VIC505). Data collection and coding were assisted greatly by Michael Allen, Margaret Beard, Ying Ching, and Sarah Fan. The authors thank all the families who have taken part in the research, and Laurence Brown and Jiansheng Guo for their comments on an earlier draft of the article. The final version has benefited greatly from anonymous reviewers' constructive and insightful comments. Correspondence concerning this article should be addressed to Sik Hung Ng at the School of Psychology, Victoria University of Wellington, P.O. Box 600, Wellington, New Zealand. Electronic mail may be sent to [email protected]
(n1.) When the analysis of variance (ANOVA) was repeated with the outlier included in the sample, exactly the same significant an d nonsignificant effects were found.
Table 1 Percentages of Single- and Multiaddressee Turns in New Zealand European and Chinese
Families, and in High- and Low-Acculturated Chinese Families
Legend for Chart: B - European (n = 12) C - Chinese Overall (n = 12) D - Chinese High Acculturated (n = 6) E - Chinese Low Acculturated (n = 6) A B C
D E Single- addressee turns 66.4[sub a] 57.3[sub a] 62.2[sub b] 52.5[sub b] Multiaddressee turns 31.6[sub c] 37.5[sub c] 31.8[sub d] 43.3[sub d] Note. Planned comparisons of means sharing the same subscript were all significant at p < .05 except for the pair of means marked "c."
Table 2 Percentages of Single- and Multiaddressee Turns Stratified According to Levels of Family
Acculturation and Generation (Chinese participants only)
Legend for Chart: B - Family Acculturation High (n = 36)
C - Family Acculturation Low (n = 35) D - Generation Youngest (n = 24) E - Generation Middle (n = 23) F - Generation Oldest (n = 24) A B C D E F Single- addressee turns 61.8[sub a] 52.5[sub a] 64.9[sub c] 58.9[sub d] 47.9[sub cd] Multi-addressee turns 31.8[sub b] 43.6[sub b] 29.0[sub e] 37.4[sub e] 46.5[sub e]
Note. Family acculturation: Means in the same row that share the same subscript are significantly different from one another at p < .05. Generation: Means in the same row that share the same subscript are significantly different from one another at p < .05.
Table 3 Summary of Hierarchical Regression Analyses for Variables Predicting Single- and
Multiaddressee Turns (N = 71)
Legend for Chart: B - R[sup 2] C - Variable(s) D - Beta E - t A B C D E
Single- addressee turns Step 1 .09 English turns .30 2.65 Step 2 .30 Generation .46 4.55 English turns .26 2.57
Multi-addressee turns Step 1 .16 English turns - .40 -3.62 Step 2 .40 Generation - .50 -5.26 English turns - .35 -3.76 Note. Delta R[sup 2] = .21 for Step 2 in single-addressee turns and .24 for Step 2 in multiaddressee turns (ps < .05). All t-values are significant at p < .05.
Ambady, N., Koo, J., Lee, F., & Rosenthal, R. (1996). More than words: Linguistic and nonlinguistic politeness in two cultures. Journal of Personality and Social Psychology, 70, 996-1011.
Bales, R. F. (1950). Interaction process analysis. Chicago: University of Chicago Press.
Bond, M., & Smith. P. B. (1996). Cross-cultural social and organizational psychology. Annual Review of Psychology, 47, 205-235.
Brown, P., & Levinson, S. C. (1987). Politeness: Some universals in language usage. Cambridge, UK: Cambridge University Press.
Button, G., & Lee, J.R.E. (1987). Transcript symbols. In G. Button & J.R.E. Lee (Eds.), Talk and social organization (pp. 9-17). Clevedon, UK: Multilingual Matters.
Chang, H-C. (1997). Language and words: Communication in the Analects of Confucius. Journal of Language and Social Psychology, 16, 107-131.
Chang, H-C. (1999). Actualizing "collective" ties through "individual" talk: Four cultural mechanisms in a Chinese society. Working paper, Department of Communication, University of Illinois at Chicago.
Clark, H. H. (1985). Language use and language users. In G. Lindzey & A. Aronson (Eds.), The handbook of social psychology (3rd ed., pp. 179-231). New York: Harper & Row.
Clyne, M. (1994). Intercultural communication at work: Cultural values in discourse. Cambridge, UK: Cambridge University Press.
Duncan, S., Jr., & Fiske, D. W. (1977). Face-to-face interaction: Research, methods, and theory. Hillsdale, NJ: Lawrence Erlbaum.
Dunne, M., & Ng, S. H. (1994). Simultaneous speech in small group conversation: All-together-now and one-at-a-time? Journal of Language & Social Psychology, 13, 45-71.
Grice, H. P. (1975). Logic and conversation. In P. Cole & J. Morgan (Eds.), Syntax and semantics: Vol. 3. Speech acts (pp. 107-142). New York: Academic Press.
Gu, Y. (1990). Politeness phenomena in modern China. Journal of Pragmatics, 14, 237-257.
Gudykunst, W. B., & Matsumoto, Y. (1996). Cross-cultural variability of communication in personal relationships. In W. B. Gudykunst, S. Ting-Toomey, & T. Nishida (Eds.), Communication in personal relationships across cultures (pp. 19-56). Thousand Oaks, CA: Sage.
Gudykunst, W. B., Ting-Toomey, S., & Nishida, T. (1996). Communication in personal relationships across cultures. Thousand Oaks, CA: Sage.
Hall, E. (1976). Beyond culture. Garden City, NY: Doubleday.
Hofstede, G. (1980). Culture's consequences: International differences in work-related values. Beverley Hills, CA: Sage.
Hollander, E. P. (1985). Leadership and power. In G. Lindzey & E. Aronson (Eds.), Handbook of social psychology (3rd ed., pp. 485- 537). New York: Random House.
Kagitcibasi, C., & Berry, J. W. (1989). Cross-cultural psychology: Current research and trends. Annual Review of Psychology, 40, 493-531.
Matsumoto, Y. (1988). Reexamination of the universality of face: Politeness phenomena in Japanese. Journal of Pragmatics, 12, 403-426.
Messick, D. M. (1988). On the limitations of cross-cultural research in social psychology. In M. H. Bond (Ed.), The cross-cultural challenge to social psychology (pp. 41-47). Newbury Park, CPC Sage.
Murata, K. (1994). Intrusive or co-operative? A cross-cultural study of interruption. Journal of Pragmatics, 21,385- 400.
Ng, S. H. (1996). Power: An essay in honor of Henri Tajfel. In W. P. Robinson (Ed.), Social identity: The developing legacy of Henri Tajfel (pp. 191- 214). Oxford, UK: Butterworth-Heinemann.
Ng, S. H., & Bradac, J. J. (1993). Power in language: Verbal communication and social influence. Newbury Park, CA: Sage.
Ng, S. H., Liu, J. H., Weatherall, A., & Loong, C.S.F. (1998). Inter-generational relationships in the Chinese community. In S. H. Ng, A. Weatherall, J. H. Liu, & C.S.F. Loong (Eds.), Ages ahead: Promoting inter -generational relationships (pp. 85-104). Wellington, New Zealand: Victoria University Press.
Ng, S. H., & Zhu, Y. (1999).Attributing causality and remembering events by collectivists and individualists in Beijing, Hong Kong and Wellington. Unpublished report, School of Psychology, Victoria University of Wellington, New Zealand.
Parker, K.C.H. (1988). Speaking turns in small group interactions: A context-sensitive event sequence model. Journal of Personality and Social Psychology, 54, 965- 971.
Reid, S. A., & Ng, S. H. (1999). Language, power and intergroup relations. Journal of Social Issues, 55, 119-139.
Roger, D., Bull, P., & Smith, S. (1988). The development of a comprehensive system for classifying interruptions. Journal of Language and Social Psychology, 7, 27-34.
Sacks, H., Schegloff, E., & Jefferson, G. (1974). A simplest systematics for the organization of turn-taking for conversation. Language, 50, 696-735.
Schachter, S. (1951). Deviation, rejection, and communication. Journal of Abnormal and Social Psychology, 46, 190-207.
Scherer, K. R. (1997). The role of culture in emotion-antecedent appraisal. Journal of Personality and Social Psychology, 73, 902-922.
Schiffrin, D. (1994). Approaches to discourse. Oxford, UK: Blackwell.
Scollon, R., & Scollon, S.B.K. (1995). Intercultural communication: A discourse approach. Cambridge, UK: Blackwell.
Shaw, M. E. (1981). Group dynamics: The psychology of small group behavior (3rd ed.). New York: McGraw-Hill.
Singelis, T. M., & Brown, W. J. (1995). Culture, self, and collectivist communication: Linking culture to individual behavior. Human Communication Research, 21, 354-389.
Singelis, T. M., Triandis, H. C., Bhawuk, D., & Gelfand, M. (1995). Horizontal and vertical dimensions of individualism
: A theoretical and measurement refinement. Cross-Cultural Research, 29, 240-
Stasser, G., & Taylor, L. A. (1991). Speaking turns in face-to-face discussions. Journal of Personality and Social Psychology, 60, 675-684.
Triandis, H. C. (1990). Cross-cultural studies of individualism
. In J. Berman (Ed.), Nebraska
Symposium on Motivation 1989 (Vol. 37, pp. 41-133). Lincoln: University of Nebraska Press.
Triandis, H. C. (1994). Culture and social behavior. New York: McGraw-Hill.
Wheeler, L., Reis, H., & Bond, M. (1989). Collectivism
in everyday social life. Journal of
Personality and Social Psychology, 57, 79- 86.
Yum, J. O. (1988). The impact of Confucianism on interpersonal relationships and communication patterns in East Asia. Communication Monograph, 55, 374-388.
By Sik Hung Ng; Cynthia S. F. Loong; An Ping He; James H. Liu and Ann Weatherall
Copyright of Journal of Language & Social Psychology
is the property of Sage Publications Inc. and its content may not be copied or
emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print,
download, or email articles for individual use. Source:
Journal of Language & Social Psychology, Mar2000, Vol. 19 Issue 1, p26, 20p. Item Number:
Editorial it time to ditch this fixation? Martin Livingston MD, FRCPsych Professor Cunningham Owens, in his paper com- cacy. The differences between drugs usually lie in side-paring older and newer antipsychotics in this issueeffect profiles, often the key factor in drug selection. (Antipsychotics: is it time to end the generation game?,see pages 48–50), has drawn several important c
Krisenprävention bezeichnet theoretische Ansätze und 14:00 Anreise, Kaffee/Tee Akteure, Instrumente, Strategien der p r a k t i s c h e M a ß n a h m e n d e r Vo r b e u g u n g , d i e Gewaltanwendung in zwischen- wie innerstaatlichen 15:00 Begrüßung und Einführung Krisenprävention – was leisten sie? Konflikten verhindern sollen. Das Konzept umfasst Uwe Trittmann, Studienlei