MEMORY, 2000, 8 (6), 365–376 Cross-cultural and gender differences in childhood amnesia Shelley MacDonald, Kimberly Uesiliana, and Harlene Hayne University of Otago, Dunedin, New Zealand In two experiments, we examined cross-cultural and gender differences in adults’ earliest memories. To do this, we asked male and female adults from three cultural backgrounds (New Zealand European, New Zealand Maori, and Asian) to describe and date their earliest personal memory. Consistent with past research, Asian adults reported significantly later memories than European adults, however this effect was due exclusively to the extremely late memories reported by Asian females. Maori adults, whose tradi- tional culture includes a strong emphasis on the past, reported significantly earlier memories than adults from the other two cultural groups. Across all three cultures, the memories reported by women contained more information than the memories reported by men. These findings support the view that the age and content of our earliest memories are influenced by a wide range of factors including our culture and our gender. These factors must be incorporated into any comprehensive theory of autobiographical memory. When adults are asked to report their earliest the age of 3; there was a sharp increase in the personal experiences, most can recall little about number of memories for events that occurred events that occurred prior to the age of 3 or 4. The beginning at the age of 4. difficulty we experience in recalling events from In an attempt to increase the reliability of the our infancy and early childhood is commonly estimates of the dates of their earliest memories, referred to as childhood amnesia. One approach Sheingold and Tenney (1982) interviewed parti- to the study of this phenomenon has been to ask cipants about an event that could be dated with adults to recall their earliest personal memories extreme accuracy. In their study, 4- to 20-year- and to indicate the date of the target event. In one olds were asked to report as much information as of the first empirical studies of this kind, Dudycha possible about the birth of a sibling. If participants and Dudycha (1933a) asked adults to identify were 3 or older at the time of the birth, they were their earliest memories and to estimate, as closely able to give a detailed account of the event, but as possible, when the event had occurred. Most of participants who were younger than 3 at the time the adults in this study indicated that their earliest provided very little information. These same basic memory was for an event that had occurred during findings have been replicated time and time again their third or fourth year of life. In another classic (Crovitz & Harvey, 1979; Crovitz & Quina- study by Waldfogel (1948), participants were Holland, 1976; Dudycha & Dudycha, 1933b, 1941; asked to record and to date all the experiences Kihlstrom & Harackiewicz, 1982; Weiland & they could recall that had occurred prior to their Steisel, 1958; Winograd & Killinger, 1983). eighth birthday. Relatively few of the experiences The findings just described provide strong evi- recalled occurred before the participants reached dence that, for many individuals, the period of Requests for reprints should be sent to Harlene Hayne, Psychology Department, University of Otago, Dunedin, New Zealand. Email firstname.lastname@example.org This research was supported by a grant from the Health Research Council of New Zealand. The authors would like to thank Seow Lim for her help in data collection and Julien Gross, Stuart McNaughton, and Carolyn Greco-Vigorito for their insightful comments on earlier drafts of this manuscript. Data collection with Maori participants was undertaken in consultation with members of the Ngai Tahu Iwi. Ó 2000 Psychology Press Ltd http://www.tandf.co.uk/journals/pp/09658211.html 366 MACDONALD, UESILIANA, HAYNE childhood amnesia ends sometime between 3 and Current empirical research on memory 4 years of age. However, there are a number of development conducted within this theoretical factors that may lead to individual differences in framework has shown that parents exhibit dif- the age of adults’ earliest memories. Usher and ferent narrative styles when discussing past events Neisser (1993), for example, asked adults to recall with their children. Two distinct styles have been their own hospitalisation, the birth of a sibling, a identified; one referred to as elaborative, high- move, or a death in the family. They found that elaborative, topic-extending, or reminiscent, and although participants could not report informa- the other referred to as low-elaborative, repeti- tion about a family funeral or a move that had tive, topic-switching, or practical (Fivush, 1991; occurred prior to the age of 4, they could report Fivush & Fromhoff, 1988; Hudson, 1990, 1993; small amounts of information about a hospitali- McCabe & Peterson, 1991; Reese & Fivush, 1993; sation or the birth of a sibling if these events Reese, Haden, & Fivush, 1993). The narratives of occurred when the participant was as young as 2 parents who exhibit a high-elaborative style are (see also Eacott & Crawley, 1998). These results characterised by long and richly detailed descrip- indicate that the absolute boundary for childhood tions of past events. In contrast, the narratives of amnesia may vary as a function of an individual’s parents who exhibit a low-elaborative style are early experiences. often very short; these low-elaborative parents In addition to factors related to the nature of provide few details and often repeat questions the original experience, individual differences in over and over in an attempt to elicit the correct the characteristics of the rememberer per se may response from their child. Although all parents also influence the age of earliest memory. Some exhibit aspects of both styles, they tend to use one studies, for example, have shown that women style over the other (Fivush & Reese, 1992). A report earlier memories than men (Dudycha & growing body of research has also shown that Dudycha, 1941; Schachtel, 1947; Waldfogel, 1948). individual differences in these parental narrative Although this gender difference is not always styles influence the child’s emerging ability to talk obtained (Kilhlstrom & Harackiewicz, 1982; about the past (Engel, 1995; Fivush, 1991; Fivush Rubin, Schulkind, & Rahhal, 1999; Wang, & Fromhoff, 1988; Hudson, 1990; McCabe & Leichtman, & White, 1998), when gender differ- Peterson, 1991; Peterson & McCabe 1994). Chil- ences occur, it is always the case that women’s dren of high-elaborative parents, for example, memories are earlier than men’s. provide longer, more well developed accounts of Similarly, there may also be differences in the personally experienced events than do children of age of earliest memory as a function of an indi- low-elaborative parents. vidual’s cultural background. In one study, for Given that children’s early autobiographical example, Mullen (1994) interviewed college memories have been shown to vary as a function students from two cultures, Asian and Caucasian of the way in which parents talk about the past American, about their childhood memories. In within a culture (Fivush & Fromhoff, 1988; Hud- four experiments, Mullen found a significant dif- son, 1990; McCabe & Peterson, 1991; Peterson & ference in the age of earliest memory between the McCabe, 1994; Reese et al., 1993), it has been two cultures; Asian adults reported significantly hypothesised that cross-cultural differences in the later memories than Caucasian adults. content, purpose, and frequency of adults’ past What are the potential mechanisms under- event narratives might lead to systematic differ- lying these individual differences in childhood ences in early autobiographical memories between amnesia? It has been hypothesised that, unlike cultures as well (Han, Leichtman, & Wang, 1998; other forms of episodic memory, auto- Leichtman, 1999; Miller & Moore, 1989; Miller, biographical memory is a social construction, Potts, Fung, Hoogstra, & Mintz, 1990; Wang, originating through experience, but elaborated, Leichtman, & White, 1998). altered, and maintained through interactions At present, there is some evidence to support with a number of important social partners the hypothesis that cross-cultural differences in (Fivush, 1991; Mullen, 1994; Nelson, 1993c). conversations about the past during childhood Within this context, Nelson (1993b, c) has pro- might lead to long-lasting cross-cultural differ- posed that conversations about the past between ences in adults’ memories of their childhood. parents and children play a central role in the Recall that when Mullen (1994) asked adults to emergence and long-term maintenance of early report their earliest memories, Asian adults autobiographical memories. reported memories that were significantly later ADULTS’ EARLIEST MEMORIES 367 than those reported by Caucasian adults. men and women from each of these groups to Recently, Mullen and Yi (1995) tape-recorded describe and date their earliest personal mem- naturally occurring conversations about the past ories. Based on the study conducted by Mullen between Korean and Caucasian mothers and their (1994), we predicted that Asian adults would 3-year-old children. These authors found that report later memories than Pakeha adults. Caucasian mother–child dyads talked about the Furthermore, given their strong cultural emphasis past three times more often than did Korean on the importance of the past, we also predicted mother–child dyads. Furthermore, Han et al. that Maori adults would report earlier memories (1998) have also shown that the past event nar- than adults from the other two cultural groups. ratives of 4- and 6-year-old Korean children are extremely sparse relative to those of Caucasian American children of the same age. Taken EXPERIMENT 1 together, these findings provide tentative support for the conclusion that the boundary for childhood Method amnesia may be influenced, at least in part, by an individual’s early narrative environment. Participants. For the present experiment, 96 The overarching goal of the present study was participants from three cultural backgrounds were to further examine cross-cultural differences in recruited by word-of-mouth from the student childhood amnesia. In particular, we were inter- population of the University of Otago. Of these, ested in whether or not a strong cultural emphasis 32 participants were Pakeha (M age = 20.50 years, on the importance of past experience would SE = .49 years), 32 were Maori (M age = 24.22 influence the age or content of adults’ earliest years, SE = 1.10 years), and 32 were Asian (29 childhood memories. New Zealand provides a Chinese, 3 Japanese, M age = 21.63 years, SE = .74 unique opportunity to address this question. years). There were 16 females and 16 males within Individuals from a number of cultural back- each ethnic group. grounds live side-by-side in the same communities while maintaining a strong sense of their own Procedure. A questionnaire, similar to the cultural identity. The native population, the one used by Mullen (1994), was developed for Maori, has lived in New Zealand for over 1000 the present experiment. In the first part of the years. Historically, richly descriptive accounts of questionnaire, participants were asked to pro- the past have played an important role in tradi- vide the following demographic information: (1) tional Maori culture (Biggs, 1970; Irwin, 1984). gender, (2) ethnicity, (3) first language, (4) birth- Although English is the dominant language in date, (5) birthplace, (6) parental ethnicity, (7) New Zealand, political changes over the past 35 parental birthplace, (8) primary caregiver during years have led to a resurgence of the native lan- childhood, (9) number of people in the house- guage and a renewed appreciation for the tradi- hold prior to age 8, (9) number of siblings, and tional culture of the Maori people, including their (10) birth order. In the second part of the ques- strong oral tradition (Metge, 1976). tionnaire, participants were asked to provide a Two major waves of immigration have brought complete description of their earliest personal individuals from around the world to New memory. Participants were specifically instructed Zealand. Approximately 150 years ago, there was a to report the first event they could recall and to large influx of European settlers; these individuals include any information that they could remem- came from a variety of Northern European coun- ber about where the event took place, who was tries including England, Scotland, and Ireland. present, what happened, and how they felt. As in New Zealanders of European descent are referred the Mullen (1994) study, no attempt was made to as Pakeha. Following the end of World War II, to verify the accuracy of the participants’ mem- New Zealand attracted a large Asian immigrant ories. In the third part of the questionnaire, par- population. More recently, individuals from a ticipants were asked to indicate their age at the number of Asian countries, including Malaysia, time of the event (in years and months) and the China, Japan, Taiwan, and Brunei have tempora- source of the original memory (i.e., personal rily settled in New Zealand to attend university. experience, family story, photographs). Partici- In the present experiments, we compared the pants were allowed as much time as they needed age and content of the earliest memories of Maori, to complete the questionnaire; most completed it Pakeha, and Asian adults. To do this, we asked within 15 minutes. 368 MACDONALD, UESILIANA, HAYNE Results Age of earliest memory. In order to examine the frequency distribution of the age of earliest Table 1 shows a summary of the demographic memory, each participant’s memory was information describing the participants in this assigned to a chronological year. Participants experiment. As shown in Table 1, all of the reporting memories between birth and 11 Pakeha and Maori participants were born in New months of age, for example, were assigned to the Zealand, and 29 of the Asian participants were 0–1 age category. Participants reporting mem- born in Asian countries (Malaysia, China, Japan, ories between 12 and 23 months of age were Taiwan, or Brunei). Although three of the Asian assigned to the 1–2 category, and so on. Figure 1 participants were born in New Zealand, the shows the percentage of memories reported by parents of two of these participants were born in all participants collapsed across gender and Asian countries. All of the Pakeha participants ethnicity as a function of age category in years. listed their parents’ ethnicity as Pakeha, and all of The largest percentage of memories were for the Asian participants listed their parents’ ethni- events that occurred between the ages of 3 and city as Asian. Of the participants who listed their 4. In fact, over half (62.5%) of the memories own ethnicity as Maori, half indicated that either reported were for events that occurred between their mother or father was not Maori. All of the the ages of 2 and 5. Pakeha and most of the Maori participants listed Figure 2 shows the mean age of earliest English as their first language. Although all of the memory expressed in months as a function of Asian participants could speak and read English gender and culture. A 2 (Gender) 6 3 (Culture) at the time of the present experiment, only four of analysis of variance (ANOVA) yielded a sig- them listed English as their first language. Finally, nificant main effect of Culture, F(2, 90) = 13.12, p 92% of the total sample indicated that their < .0001. Post-hoc t-tests (p < .05) revealed that all mother had been the primary caregiver during three cultures differed from one another. Overall, childhood. Asian participants reported the latest memories (M = 57.84 months, SE = 5.64) and Maori parti- cipants reported the earliest memories (M = 32.62 months, SE = 1.66). Pakeha participants reported TABLE 1 memories intermediate between these two A demographic description of the participants in Experiment 1 extremes (M = 42.88 months, SE = 2.96). There was also a main effect of Gender, F(1, 90) = 6.09, p Cultural background Variable Asian Pakeha Maori < .02 and a significant Gender by Culture inter- action, F(2, 90) = 6.72, p < .002. To evaluate this Birthplace interaction, post-hoc t-tests comparing males and New Zealand 3 32 32 females were calculated within each culture. This Asia 29 0 0 analysis revealed that the only significant gender Mother’s ethnicity Asian 32 0 0 difference in the age of earliest memory occurred Pakeha 0 32 9 for the Asian participants. As shown in Figure 2, Maori 0 0 23 Asian females reported much later memories (M Father’s ethnicity = 73.3 months, SE = 8.45) than Asian males (M = Asian 32 0 0 42.4 months, SE = 5.30). Pakeha 0 32 4 Maoria 0 0 26 Recall that within the Maori group, approxi- First language mately half of the 32 participants indicated that English 4 32 30 their mother or father was not Maori. There was Other 28 0 2 no difference in the age of earliest memory for Primary caregiver(s )b adults with one (M = 33.06 months, SE = 2.06) Mother 27 31 30 Father 10 8 9 versus two (M = 32.19 months, SE = 2.67) Maori Grandparent 5 0 0 parents. For the subgroup with only one Maori Other 2 0 2 parent, however, participants whose mother was a Maori (M = 28.17 months, SE = 4.09) reported The remaining two fathers of Maori participants were earlier memories than participants whose father Niuean and Fijian. b The total number of primary caregivers for each ethnic was Maori (M = 36.00 months, SE = 1.79). Despite group may exceed 32; some participants listed more than one the small size of the subgroups (mother only = 6; individual as their primary caregiver. father only = 9), the difference between them ADULTS’ EARLIEST MEMORIES 369 Figure 1. The frequency of adults’ earliest memories as a function of childhood age in years. The data have been collapsed across both culture and gender. approached conventional levels of significance, source(s) of his or her earliest memory. Across all t(13) = 2.02, p < .06. three cultures, most of the participants (86%) indicated that their earliest memory was based on Source of earliest memory. As described their personal recollection of their own experi- earlier, we asked each participant to list the ence1 . Within this group, however, some partici- pants also indicated that their memory had been augmented by photographs or family stories. Interestingly, of the participants who indicated that a family story was an important source of their memory (either alone or in conjunction with personal experience), 64% were Maori. The finding that Maori participants were more likely to use family story as a source of informa- tion about their earliest memory raises the possi- bility that the Maori advantage in the age of earliest memory reported earlier was due to their greater reliance on second-hand information relative to personal recollection of the same event. 1 Even if we exclude the data from participants who reported no first-hand knowledge of their earliest memory, the pattern of results remains unchanged. For the age of earliest memory, there is a main effect of Culture, F (2, 79) = 16.84, p < .0001, and a Culture 6 Gender interaction, F (2, 79) = 5.18, p Figure 2. The mean age (+1 SE) of adults’ earliest memory as < .01; for the content of these early memories, there is a main a function of gender and culture. effect of Gender, F (1, 79) = 7.62, p < .01. 370 MACDONALD, UESILIANA, HAYNE However, within the Maori group, there was no difference in the age of earliest memory for the Maori participants who used family stories as a source of information (M = 33.62 months, SE = 1.72) and those who did not (M = 32.29, months SE = 1.72). In a similar vein, it is also possible that indivi- dual differences in willingness to rely on family story may have contributed to the gender differ- ence in the age of earliest memory within the Asian sample. That is, perhaps the very late age of earliest memory among the Asian women was due to their unwillingness to use family story (or photos) as a source of information. Again, how- ever, the data do not support this conclusion. Within the Asian sample, the same number of Figure 3. The mean amount of information (+1 SE) included males and females relied on photos to augment in the adults’ written descriptions of their earliest memories as their memories (n = 3 each), and more females (n a function of gender and culture. = 2) than males (n = 0) relied on family story. between the amount of information reported and Amount of information reported. To assess the age of the memory in months either within a potential differences in the amount of information culture (Asian: r = .06, p < .73; Pakeha: r = .003, p provided, the written descriptions of the mem- < .98; Maori: r = 7.11, p < .54) or across all three ories were coded in a manner similar to that used cultural groups (r = .02, p < .85). in previous prospective studies of children’s autobiographical memory (Butler, Gross, & Emotional content of information. Overall, Hayne, 1995; Fivush, Gray, & Fromhoff, 1987; there was no apparent pattern in the emotional MacDonald & Hayne, 1996). To do this, one coder content of memories that adults recalled from assigned each participant a point for each piece of their childhood. Across both culture and gender, information that he or she provided about his or the target events ranged from potentially embar- her earliest memory. Participants were assigned rassing or traumatic experiences to experiences points for providing information about who was that, to us, appeared amazingly benign. Although present, what happened, where the event took some memories included vivid imagery and place, and how they felt at the time. Participants description, the importance of the event per se was were also assigned points for any additional not always clear. In fact, many of the events descriptive information about people, objects, and reported by the participants in our study seemed weather. Participants could earn more than a relatively unimportant. Prospective studies of single point within each of these coding children’s memory development have shown that categories. A second observer independently what children recall is different from what adults coded 25% of the memory descriptions. Inter- might consider to be the highlights of the same rater reliability for the total amount of informa- event (Hudson, 1986; Nelson, 1993a). Given this, tion reported was 93%. it is probably impossible to predict which child- Figure 3 shows the amount of information hood events will and will not become part of a contained in the participants’ written descriptions person’s permanent autobiography. of their memories as a function of both gender and culture. A 2 (Gender) 6 3 (Culture) ANOVA Family factors. There were a number of cross- yielded a significant main effect of Gender, F(2, cultural differences in the household structure of 90) = 8.73, p < .01. Overall, the narrative descrip- tions provided by females (M = 14.94, SE = 1.10) 2 contained more total information than the narra- In light of the extremely late memories reported by the tive descriptions provided by males (M = 10.96, SE Asian females, we conducted an additional 2 (Gender) 6 2 (Culture) analysis of variance on the total amount of infor- = .78). 2 There was no significant main effect of mation reported that excluded the Asian sample. Even with Culture and no significant Gender by Culture these data removed, there was still a significant main effect of interaction. Surprisingly, there was no correlation Gender, F (1, 62) = 9.91, p < .005. ADULTS’ EARLIEST MEMORIES 371 the participants in the present experiment. Over- after the age of 5 (see Figure 1), 12 of those all, Pakeha participants had fewer siblings, F(2, memories were reported by Asian participants 93) = 3.03, p < .05, were more likely to be first and 10 of those memories were reported by Asian born, F(2, 93) = 3.44, p < .05, and had fewer total females. In fact, only three Asian females repor- household members, F(2, 93) = 8.07, p < .001 than ted memories for events that occurred prior to the either the Asian or Maori participants. These age of 5. latter two groups did not differ from one another At least three factors may have contributed to on any of these measures. the discrepancy between our findings and those Past research has shown that there is a previously reported by Mullen (1994). First, it is relation between a number of family factors and possible that our findings may have been idio- the age of earliest memory. In Mullen’s (1994) syncratic to the sample of participants that we study, for example, first-born participants interviewed and may not reflect a real, replicable reported significantly earlier memories than par- difference between our results and Mullen’s. ticipants with older siblings. Furthermore, there Second, individual differences in participants’ was also a trend in her study for participants fluency with English may have contributed to the with fewer total siblings to report earlier mem- pattern of results. Recall that only four of the ories than participants with more siblings (see Asian participants in our sample listed English as also Wang et al., 1998). To assess the potential their first language. This finding raises the possi- effect of family factors on the age and content of bility that a mismatch between the predominant the memories reported in the present experi- language at the time of the original event ment, we conducted separate regression analyses (encoding) and the language at the time of the using birth order, number of siblings, and total questionnaire (retrieval) may have led to later number of household members as predictors of memories (Otoya, 1988). Although it is not clear the age of earliest memory and as predictors of how this account would result in systematic dif- the total amount of information contained in the ferences between males and females, we cannot written description. The results of these analyses rule out the possibility that language may have indicated that there was no significant relation been important. Finally, the discrepancy between between any family factor and the age of earliest our findings and those reported by Mullen may memory either within a culture or when the data reflect real differences between two different were collapsed across the entire sample. Further- Asian cultures. In the experiment conducted by more, there was also no significant relation Mullen in which the actual ethnicity of the parti- between any family factor and the total amount cipants was given, all participants were Korean. In of information reported. It is possible that our the present sample, however, the majority of the sample size was not large enough to detect the Asian participants were Chinese. This finding effect of individual family factors (i.e., birth raises the possibility that results obtained in one order, number of siblings, and total household Asian culture may not necessarily generalise to members) on the age or content of adults’ another (Han et al., 1998). earliest memories (cf. Mullen, 1994, Study 4). It With these issues in mind, the purpose of is also possible, however, that additional family Experiment 2 was twofold. First, we attempted to factors that we did not measure, such as birth replicate the finding that Chinese females spacing, might also moderate the effects of a sin- reported later childhood memories than Chinese gle factor considered in isolation. males. Second, we attempted to eliminate the effects of individual differences in fluency with English by giving all participants the opportunity Discussion to read the questionnaire and to provide their answers in Mandarin. Consistent with past research (Mullen, 1994), the results of Experiment 1 demonstrated that Asian participants reported significantly later memories EXPERIMENT 2 than European (Pakeha) participants. The later age for first memories reported by the Asians in Method our sample, however, was due primarily to the extremely late age of earliest memory reported by Participants. For the present experiment, 32 the Asian females. Of the 13 memories reported Chinese participants (16 males, 16 females, M age 372 MACDONALD, UESILIANA, HAYNE = 22.38 years, SE = .40) were recruited by word-of- hand, participants had the option of responding in mouth from the student population of the Uni- Mandarin. Of the 32 participants, 20 chose the versity of Otago. Mandarin option. To assess the effect of this manipulation, we compared the age and content Procedure. A questionnaire, identical to the of the Asian participants’ earliest memories as a one used in Experiment 1 was administered to function of the language in which they answered each participant. There were two versions of the the questionnaire. Separate one-way ANOVAs questionnaire; one version was in English, and the yielded no effect of language on either the age of other was in Mandarin. Participants were inter- earliest memory (English: M = 45.21 months, SE = viewed individually by a Chinese research assis- 6.2; Mandarin: M = 47.9 months, SE = 4.6) or on tant who was fluent in both languages. At the the amount of information reported (English: M = beginning of the interview, each participant was 12.33, SE = 1.57; Mandarin: M = 17.5, SE = 2.26). asked whether he or she would like to proceed in English or in Mandarin; 12 of the participants selected the English version and 20 participants Discussion selected the Mandarin version. For those partici- pants who answered in Mandarin, the research The results of Experiments 1 and 2 yielded two assistant translated their written answers into important findings regarding our Asian partici- English prior to coding. pants. First, our findings and those obtained by Mullen indicate that cross-cultural differences in the age of earliest memories cannot be attributed Results solely to differences in fluency with English. In both our study and in Mullen’s, the same pattern In the present sample, all of the participants were of findings was obtained whether participants born in Malaysia and all had parents whose were interviewed in their native language or in ethnicity was Chinese. Furthermore, none of the English. participants listed English as their first language; Second, although most Asian countries share a 25 participants indicated that their first language similar interdependent social structure (Markus & was Mandarin. Kitayama, 1991), our findings and those of others (Han et al., 1998; Mullen, 1994) clearly show that Age of earliest memory. As in Experiment 1, conclusions based on studies with one Asian cul- the age of earliest memory was expressed in ture do not necessarily apply to all others. In both months. These data were subjected to a one-way experiments of the present study, for example, analysis of variance across gender. Consistent with Chinese females reported significantly later the findings from Experiment 1, females (M = memories than Chinese males. In Mullen’s (1994, 54.12 months, SE = 5.21) reported memories that Study 4) Korean sample, the other hand, this was were significantly later than those reported by not the case. Furthermore, in a study of children’s males (M = 39.62 months, SE = 4.60), F(1, 30) = autobiographical memory, Han et al. have shown 4.35, p < .05. that although Korean and Chinese children have some characteristics in common, there are also Amount of information. As in Experiment 1, important differences in the amount and emo- the written descriptions of the memories were tional content of the information they report. coded for content. Participants were given a single point for each piece of information provided about the target event. A one-way analysis of GENERAL DISCUSSION variance yielded a significant effect of Gender, F(1, 30) = 4.22, p < .05. As in Experiment 1, The results of the present experiments add to a females (M = 18.50, SE = 2.70) reported more growing body of research highlighting the rele- total information than males (M = 12.31, SE = vance of culture and gender for the age and con- 1.34). tent of adults’ earliest autobiographical memories. Consistent with our original hypotheses, Maori Native language option. Recall that in adults reported significantly earlier memories Experiment 1 all participants were required to than Pakeha or Asian adults, and Asian female respond in English. In Experiment 2, on the other adults reported significantly later memories than ADULTS’ EARLIEST MEMORIES 373 Pakeha adults. Inconsistent with our hypotheses, participants indicated that family stories were an however, the age of the memories reported by important source of information for their earliest Asian male adults was virtually identical to that of memory. the memories reported by Pakeha adults. In addition, of the Maori participants with Furthermore, across all three cultures, the only one Maori parent, participants whose accounts provided by women contained more mother was Maori had a tendency to report total information than those provided by men. earlier memories than participants whose father At least three factors may have contributed to was Maori. Past research with American Cauca- these individual differences in adults’ earliest sian families has shown that mothers in parti- memories. First, individual differences in adults’ cular may play an important role in the memories may begin to unfold in the course of development of children’s early narrative skill conversations about the past with other family (Haden, Haine, & Fivush, 1997). Furthermore, members during childhood (Mullen, 1994; Wang in a wide range of cultures, women are generally et al., 1998). As described earlier, conversations more likely than men to maintain their heritage about the past between parents and children are culture and to encourage children to accept the thought to play a central role in the emergence heritage culture as well (Diaz-Guerrero, 1982; and long-term maintenance of autobiographical Eisenberg, 1986). These findings raise the possi- memories (Nelson, 1993c). From a cross-cultural bility that, in an attempt to preserve the impor- perspective, there are documented differences in tance of oral tradition within their family, Maori the content of past-event narratives directed spe- mothers may have been more motivated than cifically at children (Heath, 1982; Markus & Maori fathers to engage in conversations about Kitayama, 1991; Miller et al., 1990; Miller, Wiley, the past with their children. Furthermore, given Fung, & Liang, 1997; Mullen & Yi, 1995; Scollon that these mothers were also the primary care- & Scollon, 1981; Snow, 1983; Wells, 1985). Addi- givers, they had ample opportunity to engage in tionally, cross-cultural differences in adults’ past- converstions about the past with their children. event narratives have been shown to influence That is, the motivation to sustain the importance cross-cultural differences in children’s ability to of the past, coupled with greater time engaged in describe their own past experiences (Miller & child care may have contributed to the differ- Moore, 1989; Miller & Sperry, 1987, 1988). As ences in the age of earliest memory within our such, cross-cultural differences in adults’ earliest Maori sample. We are currently exploring memories may have their foundations in the parent–child conversations about the past in social context of parent–child discussions about Maori families. the past. That is, the way in which these early Similarly, it is possible that gender-related dif- conversations emerge shapes the content and ferences in the content of adults’ early memories complexity of the initial chapters of an indivi- may also have their roots in early childhood con- dual’s autobiography. versations about the past. In the present study, for Consistent with the findings just described, example, the narrative descriptions provided by cross-cultural differences in conversations about females contained significantly more information the past during childhood may have provided the than those provided by males. Past research with cornerstone for some of the cross-cultural differ- American Caucasion families has shown that both ences observed in the present study. For example, mothers and fathers are more elaborative during Maori adults reported significantly earlier mem- conversations about the past with their daughters ories than participants from the other two cultural than they are with their sons (Reese & Fivush, groups. The cultural relevance of past experience 1993). In turn, girls provide longer and more has been well documented for the Maori. The detailed accounts of their own past experiences traditional saying that ‘‘A Maori walks backwards (see also Han et al., 1998). On the basis of these into the future facing the past’’ (Irwin, 1984, p. 63), findings, Reese and Fivush (1993) predicted that reflects the importance of both personal and tribal girls are taught to value the importance of remi- history that characterises Maori culture. We niscing and eventually grow up to produce more hypothesise that this cultural emphasis on the past elaborative personal narratives than boys. The may result in a home environment in which talk present findings are consistent with their pre- about past experiences is actively encouraged. In diction. support of this hypothesis, a greater proportion of In addition to conversations about the past, it Maori participants relative to Asian or Pakeha is possible that differences in socialisation prac- 374 MACDONALD, UESILIANA, HAYNE tices more generally may also contribute to indi- ciated with it. From this perspective, children vidual differences in adults’ earliest memories. who grow up within the Maori community are For example, within our predominantly Chinese exposed to events that, by their very nature, may sample, males consistently reported earlier mem- be more memorable. ories than females. Like most countries in the The inclusion of Tangi in the Maori partici- world, gender differences in the socialisation of pants’ earliest personal memories also has male and female children has had a long history important implications for the generality of the in China (Clayre, 1985). While other countries findings previously reported by Usher and Neis- have moved towards greater equality between ser (1993). Recall that Usher and Neisser found boys and girls, the single-child policy adopted in that when adults were questioned about events China in 1973 has only reinforced the greater that had occurred when they were children, they economic and social value of male children that could not remember a family funeral if it had has persisted for centuries. It has been argued occurred before the age of 4 years. In the pre- that greater family emphasis on the personal sent study, however, the Tangi reported by the experiences and accomplishments of sons rela- Maori participants had occurred when these tive to daughters has contributed to gender dif- individuals were only 3. This finding suggests ferences in the children’s social and emotional that, when they are active participants in fun- development. Chinese boys, for example, score erals, some individuals can recall a death in the higher than Chinese girls on standardised family that occurred very early in childhood. The measures of self-concept (Watkins, Dong, & Xia, discrepancy between our findings and those of 1997). At least one current theory of childhood Usher and Neisser once again underscores the amnesia would predict that individual differ- importance of the larger cultural context in con- ences in self-concept would contribute to indivi- clusions regarding adults’ earliest recollections. dual differences in autobiographical memory in Telling stories about ourselves and our per- general and childhood amnesia in particular sonal experiences appears to be a universal (Bruner, 1996; Fivush, 1994; Howe & Courage, human phenomenon. Through these stories, we 1993; Mullen, 1994; Neisser, 1994). The finding gain a sense of who we are and our place in our that Chinese males reported earlier memories family and our community. When and how do than Chinese females is consistent with such a we begin to construct our life story? The results prediction. of the present study demonstrate that individual Finally, it is also possible that individual dif- differences in both culture and gender contribute ferences in adults’ early childhood memories to individual differences in adults’ earliest recol- may be due, at least in part, to individual differ- lections. Furthermore, it is likely that these indi- ences in the nature of children’s early experi- vidual differences may be minimised or ences (see also Eacott & Crawley, 1998; Usher enhanced by the greater historical context in & Neisser, 1993). In Experiment 1, for example, which they occur. At present, no single theory several Maori adults provided an account of a can easily account for all of the cross-cultural Tangi (funeral). In contrast, funerals were never and gender differences that have been reported. mentioned by the Pakeha or Asian adults. This The next step towards understanding these dif- difference may reflect cross-cultural differences ferences will be to pursue prospective studies of in the nature of the funeral experience. During autobiographical memory in boys and girls from the traditional Maori Tangi, for example, mem- different cultural backgrounds (Han et al., 1998; bers of the immediate and the extended family Mullen & Yi, 1995). In our own work in this take up residence on the Marae (meeting place). area, we are examining if and how the Maori Both children and adults sleep in the meeting oral tradition is reflected in the personal anec- house and keep watch over the dead body for dotes that are shared with children in traditional several days and nights. The emotional content Maori families. of the Tangi ritual ranges from extreme sadness to periods of great celebration. In contrast, tradi- Manuscript received 10 July 1999 tional Pakeha funerals in New Zealand usually Manuscript accepted 11 March 2000 take place on a single day, and the mourning process is punctuated by extreme sadness. Chil- dren may or may not be involved in the funeral ceremony or in any of the other activities asso- ADULTS’ EARLIEST MEMORIES 375 REFERENCES reminiscing across the preschool years. Develop- mental Psychology, 33, 295–307. Biggs, B. (1970). The Maori language past and present. Han, J.J., Leichtman, M., & Wang, Q. (1998). Auto- In E. Schwimmer (Ed.), The Maori people in the biographical memory in Korean, Chinese, and 1960s: A symposium (pp. 65–84). Auckland, NZ: B. American children. Developmental Psychology, 34, & J. Paul. 701–713. Bruner, J. (1996). The culture of education. Cambridge, Hanks, P., Long, T.H., & Urdang, L. (Eds.) (1983). MA: Harvard University Press. Collins dictionary of the English language. London: Butler, S., Gross, J., & Hayne, H. (1995). The effect of William Collins Sons & Co. Ltd. drawing on memory performance in young children. Heath, S.B. (1982). What no bedtime story means: Developmental Psychology, 31, 597–608. Narrative skills at home and at school. Language in Clayre, A. (1985). The heart of the dragon. Boston: Society, 11, 49–76. Houghton Mifflin Company. Howe, M.L., & Courage, M.L. (1993). On resolving the Crovitz, H.F., & Harvey, M.T. (1979). Early childhood enigma of infantile amnesia. Psychological Bulletin, amnesia: A quantitative study with implications for 113, 305–326. the study of retrograde amnesia after brain injury. Hudson, J.A. (1986). Memories are made of this: Gen- Cortex, 15, 331–335. eral event knowledge and the development of auto- Crovitz, H.F., & Quina-Holland, K. (1976). Proportion of biographic memory. In K. Nelson (Ed.), Event episodic memories from early childhood by years of knowledge: Structure and function in development age. Bulletin of the Psychonomic Society, 4, 517–518. (pp. 97–118). Hillsdale, NJ: Lawrence Erlbaum Diaz-Guerrero, R. (1982). The psychology of the Associates Inc. historic-sociocultural premise, I. Spanish Language Hudson, J.A. (1990). The emergence of auto- Psychology, 2, 383–410. biographical memory in mother–child conversation. Dudycha, G.J., & Dudycha, M.M. (1933a). Adolescents’ In R. Fivush & J.A. Hudson (Eds.), Knowing and memories of preschool experiences. Journal of remembering in young children (pp. 166–196). New Genetic Psychology: General, 42, 468–480. York: Cambridge University Press. Dudycha, G.J., & Dudycha, M.M. (1933b). Some factors Hudson, J.A. (1993). Reminiscing with mothers and and characteristics of childhood memories. Child others: Autobiographical memory in young 2-year- Development, 4, 265–278. olds. Journal of Narrative and Life History, 3, 1–32. Dudycha, G.J., & Dudycha, M.M. (1941). Childhood Irwin, J. (1984). An introduction to Maori religion memories: A review of the literature. Psychological (pp. 4–31). Bedford Park, South Australia: Aus- Review, 38, 668–682. tralian Association for the Study of Religions. Eacott, M.J., & Crawley, R.A. (1998). The offset of Kihlstrom, J.F., & Harackiewicz, J.M. (1982). The childhood amnesia: Memory for events that occurred earliest recollection: A new survey. Journal of Per- before age 3. Journal of Experimental Psychology: sonality, 50, 134–148. General, 127, 22–33. Leichtman, M.D. (1999). Cultural, social and matura- Eisenberg, N. (1986). Altruistic emotion, cognition, and tional influences on childhood amnesia. In L. Balter behavior. Hillsdale, NJ: Lawrence Erlbaum Associ- & C. Tamis-LeMonda (Eds.), Child psychology: A ates Inc. handbook of contemporary issues. (pp. 447–466). Engel, S. (1995). The stories children tell: Making sense New York: Garland Press. of the narratives of childhood. New York: W.H. MacDonald, S., & Hayne, H. (1996). Child-initiated Freeman. conversations about the past and memory perfor- Fivush, R. (1991). The social construction of personal mance by preschoolers. Cognitive Development, 11, narratives. Merrill-Palmer Quarterly, 37, 59–82. 421–442. Fivush, R. (1994). Constructing narrative, emotions, Markus, H.R., & Kitayama, S. (1991). Culture and self: and self in parent–child conversations about the past. Implications for cognition, emotion, and motivation. In U. Neisser & R. Fivush (Eds.), The remembering Psychological Review, 98, 224–253. self: Construction and accuracy in the self-narrative McCabe, A., & Peterson, C. (1991). Getting the story: A (pp. 136–157). New York: Cambridge University longitudinal study of parental styles of eliciting nar- Press. rative and the development of narrative skill. In A. Fivush, R., & Fromhoff, F.A. (1988). Style and structure McCabe & C. Peterson (Eds.), Developing narrative in mother–child conversations about the past. Dis- structure (pp. 217–253). Hillsdale, NJ: Lawrence course Processes, 11, 337–355. Erlbaum Associates Inc. Fivish, R., Gray, J.T., & Fromhoff, F.A. (1987). Two- Metge, J. (1976). The Maoris of New Zealand Rautahi. year-olds talk about the past. Cognitive Develop- London: Routledge & Kegan Paul. ment, 2, 393–409. Miller, P.J., & Moore, B.B. (1989). Narrative conjunc- Fivush, R., & Reese, E. (1992). The social construction tions of caregiver and child: A comparative per- of autobiographical memory. In M.A. Conway, D.C. spective on socialisation through stories. Ethos, 17, Rubin, H. Spinnler, & W.A. Wagenaar (Eds.), 43–64. Theoretical perspectives on autobiographical Miller, P.J., Potts, R., Fung, H., Hoogstra, L., & Mintz, memory (pp. 115–132). The Netherlands: Kluwer J. (1990). Narrative practices and the social con- Academic Publishers. struction of self in childhood. American Ethnologist, Haden, C.A., Haine, R.A., & Fivush, R. (1997). 17, 292–311. Developing narrative structure in parent–child Miller, P.J., & Sperry, L.L. (1987). The socialisation of 376 MACDONALD, UESILIANA, HAYNE anger and aggression. Merrill-Palmer Quarterly, 33, Reese, E., Haden, C.A., & Fivush, R. (1993). Mother– 1–31. child conversations about the past: Relationships of Miller, P.J., & Sperry, L.L. (1988). Early talk about the style and memory over time. Cognitive Development, past: The origins of conversational stories of per- 8, 403–430. sonal experience. Journal of Child Language, 15, Rubin, D.C., Schulkind, M.D., & Rahhal, T.A. (1999). 293–315. A study of gender differences in autobiographical Miller, P.J., Wiley, A.R., Fung, H., & Liang, C. (1997). memory: Broken down by age and sex. Journal of Personal storytelling as a medium of socialisation in Adult Development, 6, 61–71. Chinese and American families. Child Development, Schachtel, E. (1947). On memory and childhood 68, 557–568. amnesia, Psychiatry, 10, 1–26. Mullen, M.K. (1994). Earliest recollections of child- Scollon, R., & Scollon, S. (1981). Narrative, literacy, and hood: A demographic analysis. Cognition, 52, 55–79. face in interethnic communication. Norwood, NJ: Mullen, M.K., & Yi, S. (1995). The cultural context of Ablex. talk about the past: Implications for the development Sheingold, K., & Tenney, Y.J. (1982). Memory for a of autobiographical memory. Cognitive Develop- salient childhood event. In U. Neisser (Ed.), Memory ment, 40, 407–419. observed (pp. 201–212). New York: Freeman. Neisser, U. (1994). Self narratives: True and false. In U. Snow, C. (1983). Literacy and language: Relationships Neisser & R. Fivush (Eds.), The remembering self: during the preschool years. Harvard Educational Construction and accuracy in the self-narrative Review, 53, 165–189. (pp. 1–18). New York: Cambridge University Press. Usher, J.N., & Neisser, U. (1993). Childhood amnesia Nelson, K. (1993a). Events, narratives, memory: What and the beginnings of memory for four early life develops? In C.A. Nelson (Ed.), Memory and affect events. Journal of Experimental Psychology: Gen- in development. The Minnesota Symposia on Child eral, 122, 155–165. Psychology (Vol. 26, pp. 1–24). Hillsdale, NJ: Law- Waldfogel, S. (1948). The frequency and affective rence Erlbaum Associates Inc. character of childhood memories. Psychological Nelson, K. (1993b). Explaining the emergence of auto- Monographs, 62, 1–39. biographical memory in early childhood. In A.F. Wang, Q., Leichtman, M.D., & White, S. (1998). Collins, S.E. Gathercole, P.A. Conway, & P.E. Childhood memory and self-description in young Morris (Eds.), Theories of memory (pp. 355–385). Chinese adults: The impact of growing up an only Hove, UK: Lawrence Erlbaum Associates Ltd. child. Cognition, 69, 73–103. Nelson, K. (1993c). The psychological and social origins Watkins, D., Dong, Q., & Xia, Y. (1997). Age and of autobiographical memory. Psychological Science, gender differences in the self-esteem of Chinese 4, 7–14. children. Journal of Social Psychology, 137, 374–379. Otoya, M.T. (1988). A study of personal memories of Weiland, I.H., & Steisel, I.M. (1958). An analysis of bilinguals: The role of culture and language in manifest content of the earliest memories of chil- memory encoding and recall. Dissertation Abstracts dren. Journal of Genetic Psychology, 92, 41–52. International, 48 (9-B), 2789. Wells, G. (1985). Language development in the pre- Peterson, C., & McCabe, A. (1994). A social inter- school years. New York: Cambridge University actionist account of developing decontextualized Press. narrative skill. Developmental Psychology, 30, 937– Winograd, E., & Killinger, W.A. Jr. (1983). Relating age 948. at encoding in early childhood to adult recall: Reese, E., & Fivush, R. (1993). Parental styles of talking Development of flashbulb memories. Journal of about the past. Developmental Psychology, 29, 596– Experimental Psychology: General, 112, 413–422. 606.