Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

MEMORY, 2008, 16 (4), 395409

Remembering talk: Individual and gender differences in


reported speech

Richard Ely
Boston University, MA, USA

Elizabeth Ryan
The Ohio State University, Columbus, OH, USA

This study examined individual and group differences in the nature and frequency of reports of past
speech in the autobiographical memories of young adults. A sample of 108 participants (60 females,
48 males) responded in writing to six memory prompts. They also completed the Five Factor Inventory
(Costa & McCrae, 1992) and the Personal Attributes Questionnaire (Spence & Helmreich, 1978). The
frequency with which participants used reported speech was correlated with agreeableness, openness, and
expressivity; however, regression analyses indicated that narrators’ gender alone was the best predictor.
Females used more reported speech than did males. The findings suggest that recollections of past speech
are an under-appreciated yet important component of autobiographical memory.

Recent work has explored how personality vari- human experience, we would expect that accounts
ables as well as gender are associated with of verbal exchanges should figure prominently in
differences in the processes and content of auto- autobiographical memories. Furthermore, de-
biographical memory (e.g., Bauer, Stennes, & scriptions of past speech, as vehicles for rich
Haight, 2003; Blagov & Singer, 2004; Roberts, evaluative commentary (Labov, 1972), may con-
Carlos, & Kashdan, 2006; Rubin, Schrauf, & tribute to the vividness with which events are
Greenberg, 2003; Rubin & Siegler, 2004). The recalled. In this study we explored the possibility
current study continues in this tradition by that individual differences in the traits of extra-
examining individual and group differences in version, openness, agreeableness, and expressiv-
an understudied component of autobiographical ity, affect ratings of autobiographical memories,
memory: reported speech. Although much atten- discourse style, and verbal aptitude, and gender
tion has been devoted to elucidating the role itself might be associated with differences in
visual imagery plays in autobiographical memory memories of past speech. We also sought to
(Rubin & Greenberg, 2003), far less has been generate a more complete picture of the place
devoted to aural imagery in general, and spoken of reported speech in autobiographical memory
language specifically (Campos & Alonso-Quec- than currently exists.
uty, 2006). This neglect is somewhat surprising in In using reported speech, a speaker makes
that language is unique to humans and, as a highly explicit reference to a speech event. Reported
social species, human interactions are largely speech is typically classified on the basis of form
mediated through language (Dunbar, 2004; To- and falls into three categories: direct, indirect,
masello, 2003). Given the role language plays in and narratised (Genette, 1988). In direct speech

Address correspondence to: Richard Ely, Department of Psychology, Boston University, 64 Cummington Street, Boston, MA
02215, USA. E-mail: rely@bu.edu

# 2008 Psychology Press, an imprint of the Taylor & Francis Group, an Informa business
http://www.psypress.com/memory DOI:10.1080/09658210801949869
396 ELY AND RYAN

(I screamed, ‘‘I hate you, get away from me’’),1 the chance (given the large number of correlations
speaker purports to represent the original utter- the authors ran), the correlation for the warmth
ance, although the validity of such a representa- subscale of extraversion was highly significant
tion is questionable (Neisser, 1981). Direct speech and is line with the authors’ contention that
is analogous to Neisser’s (1981) notion of verba- ‘‘people who spend more time in conversation
tim recall. Indirect speech (. . . he said he got a and who find it more enjoyable should have more
knife with him) is similar to what Neisser (1981) conversation as part of their autobiographical
has referred to as gist, where the sense of what memories’’ (Rubin & Siegler, 2004, p. 920).
was once said is expressed in different words. Explanations for the other two correlations are
Finally, narratised speech (Matt cracked a joke admittedly more speculative, but could be indir-
about my game ability) describes a speech event ectly related to the degree to which openness and
in summary form, often through the use of a wide agreeableness are associated with dispositions
array of speech verbs and nouns that categorise towards interpersonal interaction and under-
the nature of the original speech event (e.g., chat, standing (Botwin, Buss, & Shackelford, 1997;
criticise, explain). Although a narratised report Jensen-Campbell & Graziano, 2001; McCrae,
may add additional information (e.g., addressee, 1996). The same reasoning holds true for the trait
topic), it masks the specific content of the original of expressivity. Expressivity is said to be a
utterance. significant part of the stereotypically female
The three forms of reported speech fall along a communal mode (Spence, 1991), where females
continuum of vividness that is, we believe, re- are seen as being focused on relationships and are
flected implicitly in the scale Rubin and collea- more integrated into family and community than
gues use to assess imagery in memories (Rubin are males (Bakan, 1966; Cross & Madson, 1997;
et al., 2003; Rubin & Siegler, 2004). As part of McAdams, 1988).
their Autobiographical Memory Questionnaire, There is a large literature focusing on the
participants are asked to rate the degree to which, relationship between affect and autobiographical
as they remember the event, ‘‘[they] or other memory (Levine & Pizarro, 2004). However, to
people are talking’’ on a scale running from ‘‘not our knowledge, little work to date has examined
at all’’ through ‘‘vaguely’’ and ‘‘distinctly’’ to ‘‘as specifically the relationship between affect and
clearly as if it were happening right now’’. We reported speech in autobiographical memories.
would argue that narratised reports (We chattered We speculated that there would be a correlation
quickly) are analogous to the lower end of the between the affect ratings of memories and the
scale, being somewhat ‘‘vague’’ in their particu- incidence of reported speech. This speculation
lars. Indirect reports (he told me not to go back) flows in part from a qualitative appreciation of
are certainly more distinct, reflecting the gist of the current data. However, this prediction is also
what was said, and direct reports (He whispered based on the memorable statement and memory
in my ear, ‘‘I love you’’), as ‘‘primary’’ sources, for speech research (reviewed below) and clinical
can feel as if they are taking place in the observation in which participants and clients
immediate (Holt, 2000). report that memorable utterances often had
Given the potential import of reported speech, long-term personal impact, impact that presum-
we wanted to explore the possible origins of ably was affectively charged (Hermans, Kempen,
individual differences in its use. Some evidence & van Loon, 1992; Knapp, Stohl, & Reardon,
for associations between individual differences in 1981).
the incidence of reported speech and personality These data point to the meta-representational
traits comes from the work of Rubin and Siegler nature of reported speech. Citing past speech is a
(2004). Specifically, they found positive correla- way of giving evaluative voice to the past verbal
tions between the degree to which participants thoughts, feelings, and attitudes of the self and,
rated the vividness of conversation in personal more importantly, others (Goldsmith & Pillemer,
memories and the traits of extraversion, agree- 1988; Keenan, MacWhinney, & Mayhew, 1977;
ableness, and openness, .28, .18, and .20, respec- Labov, 1972; MacWhinney, Keenan, & Reinke,
tively, all significant at the pB.05 level (N118). 1982). As the examples in the Discussion section
Although these findings could have been due to demonstrate, these representations often embody
critical judgements and, as such, appear to be
1
Unless otherwise noted, examples come from the data associated with a greater degree of affect. This
gathered for the current study. interpretation is supported by the high positive
REMEMBERING TALK 397

correlation, r(50).62, Rubin and colleagues (Crawford, Herrmann, Holdsworth, Randall, &
reported between the vividness ratings of the Robbins, 1989; Loftus, Banaji, Schooler, & Foster,
presence of talk and the intensity of felt emotions 1987). Indirect support for these claims of greater
in autobiographical memories, a correlation com- accuracy comes from a series of studies focused
parable to that between visual imagery and felt on child and parentchild discourse (Ely, Glea-
emotions (Rubin et al., 2003, Table 3, p. 891). son, & McCabe, 1996; Ely, Gleason, Narasimhan,
In addition to assessing reported speech in & McCabe, 1995; Ely & McCabe, 1993; for a
autobiographical memories, we examined partici- review see Ely & McCabe, 1996). This work
pants’ propensity to produce reported speech in found that females report more past speech than
the absence of any memory load by assessing the do males, and when using reported speech fe-
use of reported speech in a Thematic Appercep- males are more likely to offer a sense of what was
tion Test (TAT) story. This allowed us to deter- actually said by using direct and indirect speech
mine if individual and group differences in more than males.
reported speech reflected a general narrative Although not empirically tested outside
style that affected all story-like discourse, or was parentchild contexts, gender differences in re-
more particular and specific to autobiographical ported speech have been hinted at in other work
memories. For example, if the salience of re- (Johnstone, 1993; Tannen, 1990). For example,
ported speech in autobiographical memories is Johnstone (1993, p. 73) maintained that ‘‘[p]eople
mediated by a generic interest in all forms of in the women’s stories . . . sit around and talk;
conversational discourse (both real and ima- people in the men’s stories are more often . . .
gined), the incidence of reported speech in TAT
silent’’. To support her claim, she provided
stories would likely be associated with that in
descriptive statistics on a corpus of 58 adult
autobiographical memories. Finally, it may be that
narratives of personal experience. Reported
memory for speech is associated with general
speech was found in 60% of men’s stories and
verbal ability. In this study, we used the verbal
73% of female stories; furthermore, the propor-
portion of the Scholastic Assessment Test (SAT)
tion of transcript lines containing reported speech
as a broad measure of verbal aptitude.
was 8% for men’s stories versus 12% for women’s
Gender differences in reported speech are
likely to be secondary to more general gender stories.2
differences in memory. There are well-documen- There are a variety of not mutually exclusive
ted differences in verbal memory tasks with theoretical perspectives that could explain gender
females consistently outperforming males, and differences in the frequency of reported speech,
this is especially true for tasks involving episodic including those that point to biological and social
memory (Herlitz, Nilsson, & Bäckman, 1997; factors. Sex differences in brain organisation
Herlitz & Yonker, 2002; Lewin, Wolgers, & (Persinger & Richard, 1995) and hormones are
Herlitz, 2001). In memory for both everyday argued to play a role in women’s superior
and dramatic experiences, females are more performance in episodic memory tasks (Jacobs
focused on interpersonal interactions than are et al., 1998; Wolf & Kirschbaum, 2002; but see
males (Niedźwieńska, 2003; Sehulster, 1995). For Yonker, Eriksson, Nilsson, & Herlitz, 2003, for
example, Buckner and Fivush (1998) found that negative evidence). From a social perspective,
girls’ personal narratives focused more on affilia- women’s traditional role as caretakers and com-
tion, while boys’ centred on mastery and perfor- municators (Valian, 1998) suggest that remembe-
mance. Similarly, in a study of young children’s ring past speech may be subordinate to remem-
personal narratives, girls included more themes of bering more about interpersonal interactions in
communion than did boys (Ely, Melzi, Hadge, & general and the language that supports such
McCabe, 1998). Women’s memories about past interactions in particular. Existing evidence in-
relationships are also more vivid than those of dicates that gender-specific socialisation is
men (Acitelli & Holmberg, 1993), and, in general,
2
women’s memory styles are characterised by Narrative length can affect the likelihood of finding a
greater specificity (Pillemer, Wink, DiDonato, & single instance of reported speech. Johnstone does not report
data on the length of the narratives she analysed. The
Sanborn, 2003).
examples she cites suggest that they were likely, on average,
In this regard it is interesting to note that longer than the individual memory reports analysed here, but
women’s self-ratings of memory accuracy for shorter than our (within-participant) pooled autobiographical
conversation are higher than those of men discourse.
398 ELY AND RYAN

pervasive, with girls more than boys being en- we anticipated that the frequency of accounts of
couraged to focus on relationships, emotions, and past speech events would be positively associated
communication in their reminiscing (Bauer et al., with the traits of extraversion, agreeableness,
2003; Ely et al., 1996; Fivush, Haden, & Reese, openness, and expressivity. We also predicted
2006; Reese, 2002). that there would be significant correlations be-
We turn now to look more broadly at the place tween affect ratings and the frequency of re-
of reported speech in autobiographical memory. ported speech. Finally, we wanted to rule out as
There is a small body of literature that has looked confounds a general interest in conversational
at memories for specific speech events. Knapp discourse style that favoured the inclusion of
and colleagues (1981) collected data on single, reported speech in all narrative texts, and general
significant, ‘‘memorable messages’’. Memorable verbal aptitude.
messages were typically rule structured (if-then) In our second goal, in line with the data on
prescriptive statements (If you want a relationship child and parentchild discourse, we sought to test
to work . . . you have to work at it, p. 31) that had the prediction that, overall, females would re-
long-term, generally positive effects on recipients. member more speech than would males. Given
Verbatim recall was rated as ‘‘very certain’’ by females tendency towards greater specificity in
71% of participants. Goldsmith and Pillemer episodic memory and our past research, we also
(1988) assessed a somewhat similar corpus of predicted that females would use more direct and
remembered statements from a range of remem- indirect speech than would males. The final goal
bered speakers (e.g., unspecified, parent, teacher, of the present study was to assess generally the
and peer). They also found a relatively high frequency and form of reported speech in the
degree of confidence in verbatim recall. Overall, autobiographical memories of young adults.
nearly half of all statements were personal in Based in part on Johnstone’s (1993) data, we
nature, either reflecting the speaker’s feelings, expected that such reports would be pervasive in
attitudes, or wishes, or his or her evaluation of the memories of past personal experiences. Based on
listener (participant). A relatively small propor- memorable statement and the memory for con-
tion of statements (between 8% and 36%) dated versation literature, we expected that direct
from more than 1 year in the past. speech would be relatively rare, and indirect
There is a much larger literature that addresses and narratised speech more common. Given the
the capacity and validity of memory for past extant data, we made no prediction about
conversation. These data, collected in both la- the relative frequencies of reports of speech of
boratory and naturalistic settings, suggest that the self or others.
recall is poor after intervals as short as 5 minutes,
with one study reporting a recollection rate as low
as 10% (Stafford & Daly, 1984). However others METHOD
(e.g., Ling & Coombe’s 2005 adult sample) report
higher rates of recall (5153%) after a period of 1 Participants
week. In general, gist recall is better than
verbatim recall (Bruck, Ceci, & Francoeur, 1999; Participants were 60 female and 48 male under-
Campos & Alonso-Quecuty, 2006; Neisser, 1981), graduates who received course credit for their
and personal and unusual comments are better participation. Their average age was 19.1 years
remembered than are impersonal and mundane (SD1.2).
comments (Keenan et al., 1977; MacWhinney
et al., 1982). Some data indicate that speakers
remember more of what they themselves have Materials
said than what their partners have said (Miller, de
Winstanley, & Carey, 1996; Wagner, 1984). How- Participants were presented with a packet that
ever, other data suggest that recollections of contained seven written prompts. One was a TAT
speech of the self are no more frequent than stimulus (Plate 3 from Smith, 1992, which depicts
recollections of speech of others (Hyman & a couple sitting on a bench looking out on a
Neisser, 1992). river). Standard TAT instructions were employed.
In this study our first goal was to explore The other six prompts, presented to all partici-
individual and group differences in reported pants in the same order, were requests to describe
speech. Based on the literature reviewed earlier, participants’ earliest memory, an early childhood
REMEMBERING TALK 399

memory, a learning experience, a recent low was .69, representing an adequate level of internal
point, a recent high point, and a self-defining reliability.
memory (Singer & Salovey, 1993). These prompts As part of a brief demographic questionnaire
were selected in order to generate a rich range of included in the packet, participants who took the
salient autobiographical memories across the life SAT were asked to report their verbal score. The
span of young adults (McAdams, 1988). In addi- concern that participants might inflate their self-
tion, participants were asked to estimate their age reported scores was mitigated by the finding that
at the time the described events took place and the mean score of participants in our study was
to rate the affective quality of the memory on a within 8 points of the mean score of entering
7-point scale running from ‘‘negative’’ (1) freshmen (619) the year the data were gathered.
through ‘‘neutral’’ (4) to ‘‘positive’’ (7).
Also in the packet was the Five Factor
Inventory (Costa & McCrae, 1992) and the Design and procedure
Personal Attributes Questionnaire (Spence &
Helmreich, 1978). The Five Factor Inventory is Participants were told they were participating in a
an abridged version of the widely employed study of memory and personality. They were
NEO-Personality Inventory (Costa & McCrae, tested in small groups in a classroom setting.
1994). It produces summary scores for each of the They were presented with the TAT prompt first,
‘‘Big Five’’ traits. In the current study we focused and then the six memory prompts in the order
exclusively on the traits of extraversion, openness, listed earlier. They were given as much time as
and agreeableness. For each trait respondents rate they needed to respond to each prompt. Partici-
12 self-descriptive statements on a 5-point scale, pants then completed the questionnaires in the
ranging from ‘‘strongly disagree’’ to ‘‘strongly order presented earlier. There were several addi-
agree’’. Examples include: ‘‘I really enjoy talking tional questionnaires in the packet that were
to people’’ (extraversion) and ‘‘I generally try to unrelated to the current study. The procedure
be thoughtful and considerate’’ (agreeableness). lasted less than 2 hours.
The Five Factor Inventory has good internal and
testretest reliability, and correlates highly with
the NEO-Personality Inventory, a measure with Data preparation
well-documented construct, content, and criterion
validity (Costa & McCrae, 1994). In the current All hand-written narrative data were transcribed
study the Cronbach alphas for extraversion, to text files. We read through all files and
agreeableness, and openness were .79, .68, and identified instances of reported speech, excluding
.75, respectively, representing adequate to good occurrences that were not explicitly applicable to
reliability. speech contexts (e.g., One large envelope stuck
The Personal Attributes Questionnaire out and it said BU [Boston University] on the top
(Spence & Helmreich, 1978) is a 24-item ques- left corner). The data were then coded for
tionnaire that assesses traits stereotypically asso- speaker identity and form. In addition, we made
ciated with masculinity and femininity. We counts of the number of words each report of
focused exclusively on the eight-item female- speech encompassed, including in the count the
stereotypic trait of expressivity. Respondents reported speaker, manner of speech (I remember
rate themselves on a 5-point scale; each item her telling me . . . jokingly . . .), speech verb or
contrasts the extremes of a trait or dimension. noun, content of speech, and addressee.
Examples include ‘‘not at all emotional . . . very
emotional’’ and ‘‘not at all understanding of
others . . . very understanding of others’’. The Coding
Personal Attributes Questionnaire has good in-
ternal consistency (Abele, 2003). Testretest re- Based on narrative content, we coded each
liability has not been reported for the 24-item reported speech event as speech of the singular
measure, although adequate testretest reliability self (So I asked if it was alright to stay with him),
was reported for the original 55-item measure plural self (We talked and he made me feel
(Lenney, 1991). The measure is also reported to better), or speech of other (. . . my mother
have good criterion validity (Spence, 1993). In the announced that she was going to adopt a baby).
current study, the Cronbach alpha for expressivity We also coded the form as direct, indirect, or
400 ELY AND RYAN

narratised. In written discourse, direct speech is r(60).52, pB.001. In order to control for
marked by the use of a speech verb (e.g., say, tell) variations in length, we computed a standardised
and is enclosed in quotation marks. Indirect rate of reported speech; the number of instances
speech is often marked linguistically by the use of reported speech per narrative was divided by
of a nominal or infinitive clause or by transposi- the narrative word count and then multiplied by
tions in person or verb tense. Finally, narratised 100. In almost all instances, analyses run on raw
speech refers summarily to a past speech event frequencies duplicated those run on standardised
using a wide range of speech verbs and nouns frequencies, but with larger effect sizes.
that, while sometimes characterising the nature of The mean standardised rate across all partici-
the speech event (e.g., argument), provides little pants’ memory reports was 1.0 (SD0.6); in
sense of what was actually said. other words, on average there was one instance
The data were coded by the first author. In of reported speech per 100 words of text. The
order to assess reliability, more than one quarter percentage of text encompassing reports of
(27.2%) of the data were randomly selected and speech across all memory reports was 8.3 (SD
coded by the second author. Both authors were 5.2) and ranged from a low of 5.7 (SD9.8) in
unaware of the sex of the participants. Cohen’s memories of a high point to 10.4 (SD12.3) in
kappas for the identity of the reported speaker memories of a learning experience (Table 1).
and form of reported speech were .92 and .79
respectively, representing ‘‘substantial’’ to ‘‘near
perfect’’ agreement (Landis & Koch, 1978). Nature of reported speech

In speaker identity, participants principally


RESULTS quoted the speech of others (55.6%). First person
(self) reports represented more than a third of all
In this section we first present our descriptive reports (36.9%), and reports of the plural or we-
data on the frequency and nature of reported self were relatively rare (7.5%). A series of paired
speech in order to provide a general sense of the t-tests indicated that the proportion of each
place of reported speech in autobiographical category of reported speaker differed signifi-
memory. We then turn to look at the findings cantly from all other categories, with all ps
regarding individual and group differences. (B.001) exceeding the Bonferroni adjusted alpha
level of .01.
In form, direct speech was relatively uncom-
Frequency of reported speech mon, constituting only 10.2% of all reports.
Indirect speech represented 30.8% of all reports,
All participants complied with our request for a with narratised accounts being the most common
TAT story. Five participants failed to respond to (59.0%). As was the case with reported speaker, a
one memory prompt, and one participant failed to second series of paired t-tests indicated that the
respond to two. In terms of pervasiveness, only six proportion of each form of reported speech
(5.6%; five males, one female) participants failed differed significantly from all others, with all ps
to produce at least one report of speech within (B.001) exceeding the Bonferroni adjusted alpha
the six memory reports. The pervasiveness of level of .01.
instances of reported speech within individual
categories of memory reports was somewhat
lower, ranging from 48.1% in narratives about a Effects of prompts
recent high point to 65.7% in narratives about
a learning experience and averaged 57.1% Paired t-tests indicated that the standardised rates
(Table 1). of reported speech in early (1.10) and learning
Women (M150.4, SD26.7) produced (1.28) memories exceeded the rate in memories
longer narratives than did men (M124.8, of a high point (.67), all ps less than the
SD32.0), t(106)4.53, pB.001, d.87, and Bonferroni adjusted alpha level of .003. Although
the length of memory reports was positively small cell sizes precluded meaningful compari-
correlated with the frequency of reported speech sons of individual coding categories between
r(108).56, pB.001. This correlation held for memory reports, it is worth noting that the
both males, r(48).41, pB.005, and females, frequency of direct speech, the most vivid form
REMEMBERING TALK 401

TABLE 1
Descriptive data by memory prompt

Earliest Early Learning Low High Defining

Mean number of words in memory 133.17 (38.04) 150.28 (42.33) 144.14 (43.72) 147.81 (38.77) 128.64 (37.33) 130.37 (51.26)
reports
Mean number of instances of 1.29 (1.74) 1.71 (1.90) 2.02 (2.48) 1.49 (1.82) 0.92 (1.37) 1.34 (2.00)
reported speech
Mean number of words 10.44 (15.04) 13.60 (14.57) 16.28 (20.24) 11.81 (14.99) 8.10 (15.35) 10.87 (16.06)
encompassing reported speech
Percentage of words encompassing 7.55 (11.26) 8.80 (9.65) 10.39 (12.34) 7.41 (8.90) 5.65 (9.83) 7.71 (11.44)
reported speech
Percentage of participants reporting 54.6 63.9 65.7 59.3 48.1 50.9
speech at least once
Mean self-reported age (years) at 4.1 (2.1) 8.9 (2.5) 15.2 (3.2) 18.1 (1.7) 18.3 (1.6) 13.9 (4.6)
time of memory
Mean standardised rate of speech of .28 (.54) .30 (.57) .52 (.80) .40 (.60) .22 (.43) .37 (.80)
self
Mean standardised rate of we speech .01 (.13) .06 (.20) .05 (.17) .09 (.28) .15 (.48) .10 (.32)
Mean standardised rate of speech of .63 (.90) .74 (.94) .71 (.98) .44 (.71) .30 (.58) .48 (.79)
others
Mean standardised rate of direct .13 (.33) .13 (.37) .12 (.35) .06 (.23) .11 (.44) .12 (.40)
speech
Mean standardised rate of indirect .32 (.62) .47 (.76) .36 (.64) .29 (.58) .16 (.35) .23 (.49)
speech
Mean standardised rate of narratised .48 (.78) .50 (.73) .80 (1.07) .59 (.85) .41 (.73) .59 (.92)
speech
Mean standardised rate of all .93 (1.18) 1.10 (1.23) 1.28 (1.48) .94 (1.09) .67 (.97) .94 (1.37)
reported speech

Standard deviations in parentheses.

of reported speech, appears unrelated to age. As zero-order correlations between the three FFI
can be seen in Table 1, the highest standardised traits of interest, expressivity, verbal SAT scores,
rates of direct speech (.13) appeared in partici- and the overall frequency of reported speech. As
pants’ earliest and early childhood memories, and can be seen in Table 2 there were significant
these reports represented their oldest memories, positive correlations between the standardised
dating back on average 10 or more years. The rate of reported speech and openness, agreeable-
lowest standardised frequency of direct speech ness, and expressivity. However, contrary to our
was in autobiographical memories about a recent expectations, extraversion was not significantly
low point (.06), an event that took place on correlated with the overall rate of reported
average only a year earlier. speech. Reported speech in TAT stories was not
Thus, in terms of our goal of describing the predictive of reported speech in autobiographical
nature and frequency of reported speech, the data memories. Finally, verbal SAT scores were not
indicate that almost all participants cited at least associated with the frequency of reported speech.
one report of speech in their autobiographical In order to determine how well some of these
memories. Reported speech comprised more than correlates predicted the frequency of reported
8% of narrative texts. The speech of others was speech, we ran a linear regression with openness,
cited more than the speech of self, and narratised agreeableness, and expressivity entered as pre-
reports were the predominate form of reported dictors. One of the zero-order correlations be-
speech. tween these three predictors was significant; there
was a positive correlation between agreeableness
and expressivity, r(108).35, pB.001. The model
Correlates of individual differences in was significant, R2 .10, adjusted R2 .08, F(3,
rates of reported speech 104)3.91, p.02, with only openness signifi-
cantly predicting the frequency of reported
In order to assess our primary goal of examining speech (Table 2). When gender was entered as
individual correlates of reported speech, we ran an additional predictor, the amount of variance
402 ELY AND RYAN

TABLE 2
Zero order and partial correlations between agreeableness, openness, expressivity, gender, and the
standardised rate of reported speech

Zero order correlation Partial correlation without gender entered Partial correlation with gender entered

Extraversion .10
Openness .22* .22* .17
Agreeableness .20* .15 .07
Expressivity .19* .13 .11
Gender .35*** .26**

N 108.
*p B.05, ** p B.01, *** p B.001.

explained increased, R2 .16, adjusted R2 .13, Gender differences in reported speech


F(4, 103)4.92, p.001. However, in this model,
gender was the only significant predictor. We also In our second goal, we had predicted that females
undertook mediational analyses for openness, would use more reported speech than would
agreeableness, and expressivity. However, the males. A multiple analysis of variance (MAN-
analyses did not go beyond Baron and Kenny’s OVA) with gender as the factor was run across all
(1986) third step in which all three possible six categories of narratives. There was a signifi-
mediators failed to affect the dependent variable. cant multivariate effect of gender, Wilks’ L.84,
Table 3 displays the correlations between the F(6, 101)3.12, pB.01, h2 .16, driven princi-
standardised rates of reported speech within pally by univariate effects of gender in three
individual memory categories and the respective categories: earliest memory, learning experience,
affective ratings of those memories. Four of the and a recent high point (Table 4). However, in all
six correlations were significant, and all but one six categories of narratives, females produced
(recent high point) were negative. In other words, more reported speech than did men. Pooling the
with the exception of the prompt that explicitly data across individual participants’ narratives,
called for the recollection of what was presum- females reported an average of 1.2 reports per
ably a positive (high point) experience, negative 100 words (SD.6) compared with an average of
affect was associated with an increase in the .8 per 100 words (SD.6) for males, t(106)
frequency of reported speech, and significantly 3.79, pB.001, d.67. There was no multivariate
so in the two memories from childhood and from effect of gender in terms of the percentage of
a recent low point. There were no gender words reporting speech across all six memory
differences in affect ratings. reports. However, pooling the total number of
words of reported speech over the total number
of word of memory reports did reveal that
TABLE 3
Correlations between affect ratings of memories and TABLE 4
standardised frequencies of reported speech Mean standardised rates of reported speech and univariate
effects of sex on frequency of reported speech across
Memory n r narrative prompts

Earliest memory 107 .27** Prompt Females Males F p


Early memory 107 .31**
Learning experience 105 .07 Earliest memory 1.1 (1.2) .7 (1.0) 4.80 .04
Recent low point 106 .23* Early memory 1.2 (1.2) 1.0 (1.3) .71 ns
Recent high point 106 .22* Learning experience 1.6 (1.5) .9 (1.4) 4.92 .03
Self defining memory 96 .13 Recent low point 1.1 (1.2) .8 (1.0) 1.64 ns
Recent high point .9 (1.1) .4 (.6) 8.17 .005
Ns vary because a small number of participants failed to Self defining memory 1.1 (1.5) .7 (1.2) 2.38 ns
respond to all memory prompts and/or some participants
failed to complete the affect rating scale for all memories. Degrees of freedom, 1, 106; standard deviations in
*p B.05, **p B.01. parentheses.
REMEMBERING TALK 403

TABLE 5
Sex differences in extraversion, openness, agreeableness, and expressivity and main effects of covariates and sex on the
standardised rate of reported speech

Sex differencesa Effects of covariates and sexb

Females Males t p Ftrait ptrait Fsex psex Eta squared

Extraversion 30.6 (7.4) 28.2 (4.9) 2.02 .05 .19 ns 13.14 B.001 .11
Agreeableness 32.0 (6.0) 27.8 (5.6) 3.65 B.001 .94 ns 10.52 .002 .09
Openness 32.0 (5.5) 29.6 (5.6) 2.20 .03 2.65 ns 11.49 .001 .10
Expressivity 24.2 (3.9) 22.6 (3.7) 2.22 .03 1.74 ns 11.80 .001 .10
SAT verbal 631.2 (69.4) 622.9 (82.5) .52 ns

a b
Degrees of freedom, 106. Degrees of freedom, 1, 105.

females (M9.3, SD4.8) devoted a greater effect of gender on the proportional frequencies
percentage of the words in their memory reports (instances of form category of reported speech
to reporting speech than did males (M7.1, over all instances of reported speech) with which
SD5.4), t(106)2.26, pB.05, d.43. men and women used distinct forms of reported
Because gender differences in reported speech speech. In essence, when they actually reported
could be due to gender differences in personality speech, men and women were indistinguishable in
traits, and because males and females did in the form they used.
fact differ significantly in their scores on extra- We turn now to the data derived from TAT
version, openness, agreeableness, and expressivity stories that served as a control. Collecting TAT
(Table 5), we ran a series of analyses of covar- stories allowed us to compare participants’ pro-
iance (ANCOVAs) with these traits entered as pensity to generate reported speech in the ab-
covariates and sex as the between-participant sence of any memory load. The standardised rates
factor. After running preliminary analyses that of reported speech were nearly identical for
confirmed the homogeneity-of-slopes assumption females (M2.3, SD1.3) and males (M2.4,
for the variables of interest, we found that in all SD1.7), t(106) .26, p.79. Similarly, the
cases, there was a main effect of sex and no effect proportion of words devoted to reporting speech
of covariate. was similar for females (M.20, SD.13) and
In order to assess the degree to which gender males (M.21, SD.18), t(106) .56, p.58.
difference held across the categories of speaker The multivariate analyses for form of reported
identity and form of reported speech, we con- speech in standardised and proportional frequen-
ducted MANOVAs with gender as the factor. For cies were also not significant. In other words,
identity of speaker, there was a multivariate males and females did not differ in any appreci-
effect of gender, Wilks’ L.86, F(3, 104)5.54, able way in the amount or type of reported
pB.001, h2 .14. As can be seen in Table 6, the speech they used in their TAT stories.
univariate effects of gender were significant in all
three categories. In short, women more than men
cited the speech of themselves (singular and TABLE 6
Mean standardised rates and univariate effects of sex on
plural) and others. However, there was no multi- speaker and form
variate effect of sex on the proportional rates
(instances of speaker category of reported speech Females Males F p
over all instances of reported speech) with which
Speaker
men and women cited the speech of themselves or I .42 (.30) .30 (.25) 4.80 .04
others. We .10 (.12) .04 (.09) 6.80 .01
The multivariate effect of gender was also They .68 (.39) .44 (.38) 10.07 .002
significant for the form of reported speech, Wilks’ Form
L.85, F(3, 104)6.18, p.001, h2 .15. Sig- Direct .16 (.19) .06 (.11) 9.15 .003
nificant univariate effects held for two of the Indirect .36 (.27) .27 (.26) 3.30 .08
Narratised .67 (.35) .45 (.37) 10.17 .002
three categories, with women using direct and
narratised speech more frequently than men Degrees of freedom, 1, 106; standard deviations in
(Table 6). However, there was no multivariate parentheses.
404 ELY AND RYAN

DISCUSSION than other speech that was found in experimental


studies of speech recall (Miller et al., 1996;
Across their autobiographical memories, almost Wagner, 1984) was not reflected here in the
all young adults include descriptions of speech autobiographical reports of young adults. We
behaviour. This finding suggests that the vocal believe that the absence of what has been termed
grooming (Dunbar, 2004) that surrounds human a generation effect is likely due to the artificial
interactions finds itself represented in memories and impersonal nature of the conversations that
about those same interactions. In our discussion were targeted in the experimental research.
we touch first on the limitations of our study Recalling contributions to discussions with stran-
before considering our general findings in terms gers about the merits of liberal arts courses
of frequency, identity of speaker, and form of (Miller et al., 1996) or the intricacies of a criminal
reported speech. We turn then to our interpreta- case (Wagner, 1984) is not the same thing as a
tion of the individual and gender differences we young man’s recollection of how he came to learn
found before concluding with some general re- that his parents were separating.
marks about the special role reported speech
plays in human memory. My first vivid memory is when my parents told
We recognise at the outset a number of me that they were separating. It was a few weeks
limitations to our study. First, our participants after my birthday and I was standing in our den
represent a narrow segment of the population* on a beautiful day. My father told me that we
were going to have a family talk. I got excited
young, primarily middle-class, college students.
and my dad asked me why I was excited. I lied
The data obtained from these participants may
and said it was because I thought he said walk.
not generalise to other populations. In addition,
The truth was because we had done nothing as
we recognise that the memory reports elicited by
a family recently. They took me into the sun-
our prompts were, to a degree, acontextual. They
room with my mom, dad, and older brother
were directed to a generic audience and, as such,
(nine years old). They then told me that my
may lack ecological validity. We acknowledge that
father was moving out.
the sharing of these same memories with different
audiences (e.g., family, friends) and through As this example makes clear, the highly
different media (cf. speech versus writing) could personal nature of what was said by another and
affect participants’ memory styles (Pasupathi, the impact it had on him is likely to be why it was
2001). At the same time, the acontextual nature so readily and ‘‘vividly’’ recalled (Keenan et al.,
of our data collection controls for the variance 1977; MacWhinney et al., 1982). The participant
that might otherwise be present in more natural does report a single instance of his own speech,
settings. Finally, we acknowledge that some of our but his primary focus appears directed towards
measures were limited. For example, the Five what his parents said. As we will discuss in more
Factor Inventory did not allow us to assess trait detail shortly, the meta-representational nature of
facets (as is the case with the NEO Personality reported speech is particularly significant when
Inventory). Likewise, our affect rating scale was what is recalled is the personally relevant meta-
rather crude in that it did not frame specifically representations of others.
the perspective from which participants should This is not to say that the speech of the self is
make their judgements and ran along a conti- unimportant. Outside of a number of pedestrian
nuum from positive to negative. reports (which was also true of speech of others),
We found that an instance of reported speech recollected speech of the self was often related to
appeared on average once every 100 words and compelling self-revelatory or self-assertive state-
that, overall, reports of speech comprised more ments. For example, a female participant’s child-
than 8% of the text of memory reports. The hood memory included a description of how she
pervasiveness and proportional data are broadly worried for months about what, in retrospect,
in line with those reported by Johnstone (1993). seemed like a minor violation of a classroom
In terms of whose speech is reported, overall protocol:
slightly more than half of all reported speech was
the speech of another speaker, while speech of . . . finally one night I burst into tears . . . and
the singular self represented more than one third told my Mom what had happened . . . I just
of all reports. The propensity to recall more self remember being so worried . . . and then the
REMEMBERING TALK 405

weight being lifted when I finally told my Mom but significant correlations for three of the four
how upset I was. personality variables of interest, confirming at
least for openness and agreeableness what Rubin
In another example, a female participant and Siegler (2004) had reported. We had argued
reported her final reaction to numerous polite that the traits of openness, agreeableness, and
but ineffective corrections of her teachers mis- expressivity were likely to reflect a heightened
pronunciation of her name: ‘‘I screamed at her, interest in interpersonal interactions, including
‘Not BJ, Beejal, with an L!’’’ These examples conversational interactions (Botwin et al., 1997;
illustrate how memory of past speech of the self Jensen-Campbell & Graziano, 2001; McCrae,
can, in a certain circumstances, be as compelling 1996; Spence, 1991). This in turn could be
and as vivid as memory of the speech of others. associated with a tendency to focus on conversa-
In form, the majority of accounts were sum- tion in autobiographical memories. In this regard,
mary accounts (narratised speech). This is not we are not sure what to make of the absence of a
unexpected given the memory for speech litera- significant correlation between extraversion and
ture reviewed earlier. We have argued that the reported speech other than to note that there is
three forms of reported speech reflect varying not a simple relationship between extraversion
degrees of vividness, and the fact that most and reported speech, despite what we might have
recollections of conversations are towards the thought at the outset.
less vivid end of the continuum is in line with The experience of affect was another domain
Rubin and colleagues’ findings for ratings for in which individual differences were evident. We
vividness of remembered talk. Their data (from would argue that memorable speech acts are
Rubin & Seigler, 2004) suggest that on a 7-point likely to be associated with a degree of affect,
scale, participants rate talk vividness (4.23) well reflecting the importance of their content (Gold-
below ratings for ‘‘see’’ (5.61) and ‘‘setting’’ smith & Pillemer, 1988; Keenan et al., 1977;
(5.96). Knapp et al., 1981; MacWhinney et al., 1982).
Nevertheless, many other reports included This conclusion is supported by the correlations
some sense of what was once said, with a small between affect ratings and the frequency of
number (slightly more than 10%) taking the form reported speech, which parallel the correlational
of direct speech. The recollection of direct reports findings reported by Rubin and colleagues (2003).
from many years in the past is in line with the In our data, with the exception of autobiographi-
high degree of participant confidence for verba- cal memories of high points, the more negative
tim recall that was described by memorable the rating, the more frequent the incidence of
statement researchers (Goldsmith & Pillemer, reported speech, significantly so in the case of
1988; Knapp et al., 1981). While being cautious three categories (earliest, early, and recent low
given the degree to which small cell sizes pre- point reports). Although these correlations may
cluded meaningful statistical comparisons, two merely reflect memories containing greater de-
patterns in direct reports deserve comment. First, grees of emotionally charged human interactions,
the frequency of direct speech seems unrelated to such interactions are likely to be mediated
age of memory, with the first and second highest through language, the most explicit vehicle for
standardised rates occurring in reports elicited by communicating affective and cognitive informa-
prompts for the earliest and an early childhood tion (Dunbar, 2004).
memory. This suggests that vivid recall of speech The gender differences we report here are
from many years in the past, though not common, generally consistent with our earlier child and
is also not highly unusual. Second, it is notable parentchild data. Most of the effect sizes were in
that the lowest rate was found in response to the the moderate to large range (Cohen, 1988). In
prompt for a memory of a recent low point, a short, women reported more speech of them-
finding that is compatible with what is known selves and others, and used direct and narratised
about how affect and affect regulation interact speech more than did men. The gender difference
with memory specificity (Williams et al., 2007). in the proportion of words containing reported
Initially we had thought that the individual speech was similar to that for transcript lines
difference in memories of reported speech might described by Johnstone (1993). Importantly, in
be associated with individual differences in per- terms of the propensity to produce reported
sonality dispositions, in line with what has been speech in narratives not dependent on episodic
found in past research. There were indeed modest memory, men and women were very similar in the
406 ELY AND RYAN

rate and form of their reported speech in TAT earliest experiences with parentchild reminiscing
stories. Thus, the gender differences appear to be differ along gender lines (Buckner & Fivush,
restricted to autobiographical memories, and are 1998, 2000). In talking about past events, parents
likely to reflect broader qualitative differences in are more elaborative and more emotional when
the content of autobiographical memories of men conversing with daughters than with sons (Fivush,
and women, with women being more focused on Berlin, Sales, Mennuti-Washburn, & Cassidy,
interpersonal (conversational) relationships than 2003). Not surprisingly, girls’ narratives of past
are men. personal experience are in turn more elaborated
Looking specifically at the gender differences and more emotionally complex than are boys’
in speaker and form suggests first that women (Buckner & Fivush, 1998; Stapley & Haviland,
overall report more speech by all speakers than 1989). It is even possible that some speech events
do men. For form, the data suggest that women are intrinsically more emotional (as we will
recall both more specific images of conversation suggest shortly) and as such are encoded more
(direct speech) in line with our parentchild data deeply by females than by males (Bloise &
and the findings of Pillemer and colleagues (2003) Johnson, 2007; Seidlitz & Diener, 1998).
and Acitelli and Holmberg (1993). The pattern In one of our longitudinal studies of parent
for indirect speech favoured women, but was just child discourse, we found that mothers and girls
shy of significance. Of course they also recalled were more focused on talking about past speech
more general images of talk (narratised speech), a events than were fathers or boys, and girls later
result we had not anticipated. As we will discuss used more reported speech than did boys in
more fully below, these findings fit broadly within talking independently about the past with an
the notion that women are more focused on experimenter (Ely et al., 1996). These data
communication and language than are men, and suggest that attention to language is part of the
subsequently are more likely to encode, retain, female role wherein women and girls are viewed
and later reconstruct descriptions of all forms of as communicators (Ely, Gleason, MacGibbon, &
talk. More importantly, when the variables that Zaretsky, 2001). Evidence from studies of early
differed by gender (Budaev, 1999; Costa, Terrac- parentchild interactions suggest that this is in
ciano, & McCrae, 2001; Spence, 1993) were fact the case, with mothers fostering communica-
entered as covariates in ANCOVAs run with tion more than fathers, especially when interac-
reported speech as the dependent variable and ting with their daughters (Fagot & Hagan, 1991).
gender as the independent variable, the effect of We turn now to the broader implications of our
gender remained. These results along with those findings. Although Bandura (1997) has down-
of our second regression analysis are rather stark played the role verbal persuasion plays in foster-
in their implication: something beyond traits and ing or impeding self-efficacy, anecdotal and
attributes that are typically associated with gen- empirical evidence suggests that, outside the
der is likely to be driving the difference in laboratory, what was once said can have lasting
attention to past speech. and profound effects (Hermans et al., 1992;
Biological explanations for why females per- Pillemer, 2001). In this regard, it is remarkable
form better on measures of episodic memory how many of the examples included in Pillemer’s
include differences in brain organisation (Per- (2001) paper on momentous events centre around
singer & Richard, 1995) and hormones (Jacobs et reports of speech. They include Tulving’s emo-
al., 1998; Wolf & Kirschbaum, 2002). Another tional reaction to a ‘‘casual comment’’ of a friend,
possible explanation focuses on volubility. the effects of an overheard ‘‘brilliant conversa-
Although women have been cast as the more tion’’ on a woman’s pioneering career choice, and
talkative gender, recent data do not support this a number of instances where compelling remarks
stereotype (Leaper & Ayres, 2007; Mehl, Vazire, made by professors had long-term impact on their
Ramı́rez-Esparza, Slatcher, & Pennebaker, 2007). student listeners.
Thus, there is no reason to believe that women We encountered a number of such instances*
should remember more talk because they talk narratives in which speech acts were described as
more than men. constituting momentous events (Pillemer, 1998,
An explanation we are partial to focuses on 2001). Many of these memorable remarks were
differential socialisation (Bauer et al., 2003; produced by speakers of higher status than the
Fivush et al., 2006; Reese, 2002). There is a (participant) listeners (Knapp et al., 1981). For
body of work which indicates that children’s example, a male participant reports in vivid direct
REMEMBERING TALK 407

speech his coach’s response to a throw he made Baron, R. M., & Kenny, D. A. (1986). The moderator
that accidentally injured another player on his mediator variable distinction in social psychological
research: Conceptual, strategic, and statistical con-
baseball team:
siderations. Journal of Personality and Social Psy-
chology, 51, 11731182.
While I was going to see if he [the other player] Bauer, P. J., Stennes, L., & Haight, J. C. (2003).
was ok the coach just turns to me and says, Representation of the inner self in autobiography:
‘‘Get out of here you little bastard, you have Women’s and men’s use of internal states language
done enough.’’ I didn’t play baseball for 5 years in personal narratives. Memory, 11, 2742.
Blagov, P. S., & Singer, J. A. (2004). Four dimensions of
after that.
self-defining memories (specificity, meaning, con-
tent, and affect) and their relationships to self-
In our final example, a female participant restraint, distress, and repressive defensiveness.
recounts what her father said in response to a Journal of Personality, 72, 481511.
homework query: Bloise, S. M., & Johnson, M. K. (2007). Memory for
emotional and neutral information: Gender and
I remember I was asking my dad a question individual differences in emotional sensitivity.
Memory, 15, 192204.
about math. And he asked me back, ‘‘What do Botwin, M. D., Buss, D. M., & Shackelford, T. K.
you think?’’ And I said I didn’t know. He said, (1997). Personality and mate preferences: Five
‘‘You can’t be that stupid.’’ After that I always factors in mate selection and marital satisfaction.
felt that I was never good enough. I could never Journal of Personality, 65, 107136.
be the best. Till today I still can’t forgive my Bruck, M., Ceci, S. J., & Francoeur, E. (1999). The
accuracy of mothers’ memories of conversations
father for hurting me the way he did. I felt like with their preschool children. Journal of Experi-
my heart was torn in pieces. I never asked him mental Psychology: Applied, 5, 89106
anything again. This is why I’m always trying to Buckner, J. P., & Fivush, R. (1998). Gender and self in
prove myself to be better. children’s autobiographical narratives. Applied Cog-
nitive Psychology, 12, 407429.
These memories cast speech acts as pivotal Buckner, J. P., & Fivush, R. (2000). Gendered themes in
family reminiscing. Memory, 8, 401412.
events, as worthy of mention as any other Budaev, S. V. (1999). Sex differences in the Big Five
behaviour, perhaps more so in that they capture personality factors: Testing an evolutionary hypoth-
vivid and explicit representations of the affect, esis. Personality and Individual Differences, 26, 801
attitude, and cognitions of the self and others. The 813.
ability to encode, retrieve, and reconstruct these Campos, L., & Alonso-Quecuty, M. L. (2006). Remem-
bering a criminal conversation: Beyond eyewitness
representations is likely to have a played a key
testimony. Memory, 14, 2736.
role in human evolution and culture (Ong, 1982; Cohen, J. (1988). Statistical power analysis for the
Rubin, 1995). And ontologically, as our data behavioral sciences. Hillsdale, NJ: Lawrence Erl-
suggest, memories of speech are likely to figure baum Associates Inc.
prominently in the life stories of many young Costa, P. T., & McCrae, R. R. (1992). The NEO Five-
adults. Factor Inventory (NEO-FFI). Odessa, FL: Psycho-
logical Assessment Resources.
Manuscript received 19 November 2007 Costa, P. T., & McCrae, R. R. (1994). The NEO-PI-R:
Manuscript accepted 25 January 2008 Professional manual. Odessa, FL: Psychological
First published online 5 March 2008 Assessment Resources.
Costa, P. T., Terracciano, A., & McCrae, R. R. (2001).
Gender differences in personality traits across
cultures: Robust and surprising findings. Journal of
REFERENCES Personality and Social Psychology, 81, 322331.
Crawford, M., Herrmann, D. J., Holdsworth, M. J.,
Abele, A. E. (2003). The dynamics of masculine-agentic Randall, E. P., & Robbins, D. (1989). Gender and
and feminine-communal traits: Findings from a beliefs about memory. British Journal of Psychology,
prospective study. Journal of Personality and Social 80, 391401.
Psychology, 85, 768776. Cross, S. E., & Madson, L. (1997). Models of the self:
Acitelli, L. K., & Holmberg, D. (1993). Reflecting on Self construals and gender. Psychological Bulletin,
relationships: The role of thoughts and memories. 122, 537.
Advances in Personal Relationships, 4, 71100. Dunbar, R. I. M. (2004). Gossip in evolutionary
Bakan, D. (1966). The duality of human existence: perspective. Review of General Psychology, 8, 100
Isolation and communion in Western man. Boston: 110.
Beacon Press. Ely, R., Gleason, J. B., MacGibbon, A., & Zaretsky, E.
Bandura, A. (1997). Self-efficacy: The exercise of (2001). Attention to language: Lessons learned at
control. New York: W. H. Freeman. the dinner table. Social Development, 10, 355373.
408 ELY AND RYAN

Ely, R., Gleason, J. B., & McCabe, A. (1996). ‘‘Why tion. Journal of Verbal Learning & Verbal Behavior,
didn’t you talk to your Mommy, Honey?’’ Gender 16, 549560.
differences in talk about past talk. Research on Knapp, M. L., Stohl, C., & Reardon, K. K. (1981).
Language and Social Interaction, 20, 725. ‘‘Memorable’’ messages. Journal of Communication,
Ely, R., Gleason, J. B., Narasimhan, B., & McCabe, A. 31(Autumn), 2741.
(1995). Family talk about talk: Mothers lead the way. Labov, W. (1972). Language in the inner city: Studies in
Discourse Processes, 19, 201218. the Black English vernacular. Philadelphia: Univer-
Ely, R., & McCabe, A. (1993). Remembered voices. sity of Pennsylvania Press.
Journal of Child Language, 20, 671696. Landis, J. R., & Koch, G. G. (1977). The measurement
Ely, R., & McCabe, A. (1996). Gender differences in of observer agreement for categorical data. Bio-
memories for speech. In P. Thompson (Ed.), Inter- metrics, 33, 159174.
national yearbook of oral history and life stories: Leaper, C., & Ayres, M. M. (2007). A meta-analytic
Memory and gender (pp. 1730). New York: Oxford review of gender variation in adults’ language use:
University Press. Talkativesness, affiliative speech, and assertive
Ely, R., Melzi, G., Hadge, L., & McCabe, A. (1998). speech. Personality and Social Psychology Review,
Being brave, being nice: Themes of agency and 11, 328363.
communion in children’s narratives. Journal of Lenney, E. (1991). Sex roles: The measurement of
Personality, 66, 257284. masculinity, femininity, and androgyny. In J. P.
Fivush, R., Berlin, L., Sales, J. M., Mennuti-Washburn, Robinson, P. R. Shaven, & L. S. Wrightsman
J., & Cassidy, J. (2003). Functions of parentchild (Eds.), Measures of personality and social psycholo-
reminiscing about emotionally negative events. gical attitudes (pp. 573660). San Diego, CA: Aca-
Memory, 11, 179192. demic Press.
Fivush, R., Haden, C. A., & Reese, E. (2006). Elabor- Levine, L. J., & Pizarro, D. A. (2004). Emotion and
ating on elaborations: Role of maternal reminiscing memory research: A grumpy overview. Social Cog-
style in cognitive and social development. Child nition, 232, 530554.
Development, 77, 15681588. Lewin, C., Wolgers, G., & Herlitz, A. (2001). Sex
Fagot, B. A., & Hagan, R. (1991). Observations of
differences favoring women in verbal and non-
parent reactions to sex-stereotyped behaviors: Age
verbal, but not in visuospatial episodic memory.
and sex effects. Child Development, 62, 617628.
Neuropsychology, 15, 164173.
Genette, G. (1988). Narrative discourse revisited.
Ling, J., & Coombie, A. (2005). Age effects in ear-
Ithaca, NY: Cornell University Press.
witness recall of a novel conversation. Perceptual
Goldsmith, L. R., & Pillemer, D. B. (1988). Memories
of statements spoken in everyday contexts. Applied and Motor Skills, 100, 774776.
Loftus, E. E., Benaji, M. R., Schooler, J. W., & Foster,
Cognitive Psychology, 2, 273286.
Herlitz, A., Nilsson, L-G., & Bäckman, L. (1997). R. A. (1987). Who remembers what? Gender
Gender differences in episodic memory. Memory differences in memory. Michigan Quarterly Review,
& Cognition, 25, 801811. 26, 6485.
Herlitz, A., & Yonker, J. E. (2002). Sex differences in MacWhinney, B., Keenan, J. M., & Reinke, P. (1982).
episodic memory: The influence of intelligence. The role of arousal in memory for conversation.
Journal of Clinical and Experimental Neuropsychol- Memory & Cognition, 10, 308317.
ogy, 24, 107114. McAdams, D. (1988). Power, intimacy and the life story:
Hermans, H. J. M., Kempen, H. J. G., & van Loon, R. J. Personological inquiries into identity. New York:
P. (1992). The dialogical self. American Psychologist, Guilford Press.
47, 2333. McCrae, R. (1996). Social consequences of experiential
Holt, E. (2000). Reporting and reacting: Concurrent openness. Psychological Bulletin, 120, 323337.
responses to reported speech. Research on Lan- Mehl, M. R., Vazire, S., Ramı́rez-Esparza, N., Slatcher,
guage and Social Interaction, 33, 425454. R. B., & Pennebaker, J. W. (2007). Are women really
Hyman, I. E., & Neisser, U. (1992). The role of the self more talkative than men? Science, 317, 82.
in recollections of a seminar. Journal of Narrative Miller, J. B., de Winstanley, P., & Carey, P. (1996).
and Life History, 2, 81103. Memory for conversation. Memory, 4, 615631.
Jacobs, D. M., Tang, M-X., Stern, Y., Sano, M., Marder, Neisser, U. (1981). John Dean’s memory: A case study.
K., Bell, K. L., et al. (1998). Cognitive function in Cognition, 9, 122.
nondemented older women who took estrogen after Niedźwieńska, A. (2003). Gender differences in vivid
menopause. Neurology, 50, 368373. memories. Sex Roles, 49, 321331.
Jensen-Campbell, L. A., & Graziano, W. G. (2001). Ong, W. J. (1982). Orality and literacy: The technologiz-
Agreeableness as a moderator of interpersonal ing of the word. New York: Metheun.
conflict. Journal of Personality, 69, 323362. Pasupathi, M. (2001). The social construction of the
Johnstone, B. (1993). Community and contest: Mid- personal past and its implications for adult devel-
western men and women creating their worlds in opment. Psychological Bulletin, 127, 651672.
conversational storytelling. In D. Tannen (Ed.), Persinger, M. A., & Richards, P. M. (1995). Women
Gender and conversational interaction (pp. 6280). reconstruct more details than men for a complex
New York: Oxford University Press. five-minute narrative: Implications for right-hemi-
Keenan, J. M., MacWhinney, B., & Mayhew, D. (1977). spheric factors in the serial memory effect. Percep-
Pragmatics in memory: A study of natural conversa- tual and Motor Skills, 80, 403410.
REMEMBERING TALK 409

Pillemer, D. B. (1998). Momentous events, vivid mem- Smith, C. P. (Ed.). (1992). Motivation and personality:
ories: How unforgettable moments help us under- Handbook of thematic content analysis. New York:
stand the meaning of our lives. Cambridge, MA: Cambridge University Press.
Harvard University Press. Spence, J. T. (1991). Do the BSRI and PAQ measure
Pillemer, D. B. (2001). Momentous events and the life the same or different concepts? Psychology of
story. Review of General Psychology, 5, 123134. Women Quarterly, 15, 141163.
Pillemer, D. B., Wink, P., DiDonato, T. E., & Sanborn, Spence, J. T. (1993). Gender-related traits and gender
R. L. (2003). Gender differences in autobiographical ideology: Evidence for a multifactorial theory.
memory styles of older adults. Memory, 11, 525532. Journal of Personality and Social Psychology, 64,
Reese, E. (2002). Social factors in the development of 623635.
autobiographical memory: The state of the art. Spence, J. T., & Helmreich, R. L. (1978). Masculinity
Social Development, 11, 124142. and femininity: Their psychological dimensions,
Roberts, J. E., Carlos, E. L., & Kashdan, T. B. (2006). correlates, and antecedents. Austin, TX: University
Impact of depressive symptoms, self-esteem and of Texas Press.
neuroticism on trajectories of over general autobio- Stafford, L., & Daly, J. A. (1984). Conversational
graphical memory over repeated trials. Cognition & memory: The effects of recall mode and memory
Emotion, 20, 383401. expectancies on remembrances of natural conversa-
Rubin, D. C. (1995). Memory in oral traditions: The tions. Human Communication Research, 10, 379
cognitive psychology of epic, ballads, and counting- 402.
Stapley, J. C., & Haviland, J. M. (1989). Beyond
out rhymes. New York: Oxford University Press.
depression: Gender differences in normal adoles-
Rubin, D. C., & Greenberg, D. L. (2003). The role of
cents’ emotional experiences. Sex Roles, 20, 295308.
narrative in recollection: A view from cognitive
Valian, V. (1998). Why so slow?: The advancement of
psychology and neuropsychology. In G. D. Fireman,
women. Cambridge, MA: MIT Press.
T. E., McVay Jr., & O. J. Flanagan (Eds.), Narrative Tannen, D. (1990). You just don’t understand: Women
and consciousness: Literature, psychology and the and men in conversation. New York: Ballantine.
brain (pp. 5385). New York: Oxford University Tomasello, M. (2003). Constructing a language: A
Press. usage-based theory of language acquisition. Cam-
Rubin, D. C., Schrauf, R. W., & Greenberg, D. L. bridge, MA: Harvard University Press.
(2003). Belief and recollection of autobiographical Wagner, W. (1984). Recognition of own and others’
memories. Memory & Cognition, 31, 887901. utterances in a natural conversation. Personality and
Rubin, D. C., & Siegler, I. C. (2004). Facets of Social Psychology Bulletin, 10, 596604.
personality and the phenomenology of autobiogra- Williams, J. M. G., Barnhofer, T., Crane, C., Hermans,
phical memory. Applied Cognitive Psychology, 18, D., Raes, F., Watkins, E., et al. (2007). Autobiogra-
913930. phical memory specificity and emotional disorder.
Sehulster, J. R. (1995). Memory styles and related Psychological Bulletin, 133, 122148.
abilities in presentation of self. American Journal Wolf, O. T., & Kirschbaum, D. (2002). Endogenous
of Psychology, 108, 6788. estradiol and testosterone levels are associated with
Seidlitz, L., & Diener, E. (1998). Sex differences in the cognitive performance in older women and men.
recall of affective experiences. Journal of Personality Hormones and Behavior, 41, 259266.
and Social Psychology, 74, 262271. Yonker, J. E., Eriksson, E., Nilsson, L-G., & Herlitz, A.
Singer, J. P., & Salovey, P. (1993). The remembered self: (2003). Sex differences in episodic memory: Minimal
Emotion and memory in personality. New York: The influence of estradiol. Brain and Cognition, 52, 231
Free Press. 238.

You might also like