Вы находитесь на странице: 1из 26

FIRST

Article LANGUAGE

First Language
XX(X) 1–26
Gender and Patterns of © The Author(s) 2010
Reprints and permission: sagepub.
Language Development in co.uk/journalsPermissions.nav
DOI: 10.1177/0142723709359241
Mother–toddler and http://fla.sagepub.com

Father–toddler Dyads

Gretchen S. Lovas
Susquehanna University

Abstract
The study examined parent, child, and dyadic gender effects in parent reports of
words and MLUs. Mothers and fathers from 113 families completed the MacArthur
Communicative Development Inventory: Toddlers when the toddlers were 1;7; half
completed a follow-up at 2;0. Child gender differences in words and MLUs increased over
time and parent gender differences decreased. Dyadic analyses revealed bidirectional
influences. At 1;7, dyadic scores for words and MLUs displayed a descending pattern
from mother–daughter, to mother–son, to father–daughter, to father–son dyads. At 2;0,
the most and fewest words were reported in mother–daughter and mother–son dyads,
respectively; and the longest and shortest MLUs in father–daughter and father–son
dyads, respectively. The data raise questions about the ‘bridge hypothesis.’ They suggest
that fathers are more likely to provide a bridge for daughters than for sons; daughters
may play an active role in eliciting this behavior.

Keywords
bridge hypothesis, emotion socialization, gender differences, gender socialization,
language development, parent–child interactions

Language development is a domain in which gender differences are already well estab-
lished. The female advantage in verbal ability is one of the few gender differences consid-
ered to be robust. Individual studies of verbal abilities do show some inconsistency (see e.g.,
Eriksson & Berglund, 1998; Holdgrafer, 1991, who came to different conclusions about the
strength of gender differences before age 2). However, when differences are found, they
favor females. Females produce sounds at an earlier age, use words sooner, develop larger

Corresponding author:
Gretchen S. Lovas, Susquehanna University, Department of Psychology, 514 University Avenue, Fisher 106,
Selinsgrove, PA 17870, USA.
email: lovas@susqu.edu

Downloaded from fla.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016


2 First Language XX(X)

vocabularies, display greater grammatical complexity, spell better, and read sooner than
males (Bornstein, Hahn, & Haynes, 2004; Donelson & Gullahorn, 1977; Halpern, 2002;
Holdgrafer, 1991; Huttenlocher et al., 1991; Stolt, Haataja, Lapinleium, & Lehtonen, 2008).
Females also produce a higher quantity of verbal communication (Bornstein et al., 2004;
Leaper & Smith, 2004), even during the first months of life (Holdgrafer, 1991). These early
gender differences are fairly small in size and tend to disappear around age 5; however,
gender differences in certain verbal abilities reappear during adolescence (see e.g., Burman,
Bitan, & Booth, 2008) and may continue not only through middle adulthood but into old age
as well (de Frias, Milsson, & Herlitz, 2006; van Hooren et al., 2007). In addition, females
display an advantage in verbal and associational fluency, findings that are particularly
robust: these differences do not disappear as thoroughly during latency as do other verbal
differences and there is more agreement that they continue across adulthood (de Frias et al.,
2006; Donelson & Gullahorn, 1977; Halari et al., 2005; Halpern, 2002). In contrast, males
display more language problems than females across the lifespan and have much higher
rates of stuttering and various types of learning disabilities (Halpern, 2002; Levy & Heller,
1992). There is a male advantage in verbal comprehension during childhood, but this differ-
ence disappears during adolescence (Kraft & Nickel, 1995). The only consistent male
advantage in language skills is in verbal analogies (Halpern, 2002; Levy & Heller, 1992).
Why these gender differences in verbal abilities exist is still a matter of some controversy.
Three major literatures, those on biological gender differences in infancy, gender socialization,
and fathers’ influence on language development, have potential relevance to this question.

Gender Differences in Neurological Maturity and Organization


One explanation for gender differences in verbal abilities is based on research suggesting
that early gender differences in neurological maturity and organization produce gender dif-
ferences in infant attention and perception (Halpern, 2002; Levy & Heller, 1992). Such
differences are thought to mediate children’s responses to parents’ verbal and nonverbal
behavior and to reinforce gender-differentiated parent interactions with children, especially
with very young infants. Underlying this argument is evidence that gender differences in
the timing and composition of hormonal cascades during early gestation create gender dif-
ferences in rates of development, such that male development is delayed relative to female
development (Breedlove, 1994; Levy & Heller, 1992; Reinisch & Sanders, 1992). This
delay is most visible in skeletal maturity and other markers of physical growth (Breedlove,
1994; Flory, 1935; Roche, 1968), but there is general consensus that brain development in
males is also delayed, resulting in infant gender differences in left hemisphere maturity and
in the lateralization and organization of function within the brain (see e.g., Friederici et al.,
2008; Hanlon, Thatcher, & Cline, 1999; Shucard, Shucard, & Thomas, 1987).1 These, in
turn, produce gender differences in perception and attention that have been well docu-
mented (see e.g., Notman & Nadelson, 1991; Reinisch & Sanders, 1992).
There are several findings in this literature that are directly relevant to language develop-
ment. Females have been shown to display better discrimination of and attention to verbal
stimuli than males, even as newborns (Friederici et al., 2008; Friedman & Jacobs, 1981).
Females also make better eye contact during the first year of life (Hittelman & Dickes, 1979;
Leeb & Rejskind, 2004) and engage in higher levels of joint attention (Olafsen et al., 2006),
both of which may facilitate their receptivity to verbal communication and to the nonverbal

Downloaded from fla.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016


Lovas 3

signals that accompany it. Finally, females are less irritable and easier to sooth as infants
(Notman & Nadelson, 1991; Silverman, 1987), another factor that may increase their ability
to focus on and respond to verbal communication. These early child gender differences in
verbal and joint attention, eye contact, and general neurological maturity are hypothesized to
differentially reinforce parents’ verbal behaviors, creating feedback loops of mutual rein-
forcement that produce increasingly complex verbal interactions between daughters and their
parents (Halpern, 2002). Gender differences in language skills may thus reflect the incremen-
tal widening and augmentation of small gender differences that already exist at birth.

Gender Socialization
There are two areas of research on gender socialization that are relevant to language
development. The first deals specifically with the gender socialization of language and
the second with the gender socialization of emotions. These two areas of research are
interrelated, theoretically, because both have implications for communicative style and
for the emotional tone and content of verbal interaction.
From research on the gender socialization of language, we know that mothers speak more
to daughters than to sons, thus providing daughters with greater exposure to language (Berry,
Poortinga, Segall, & Dasen, 1992; Clearfield & Nelson, 2006; Golombok & Fivush, 1994;
Leaper, Anderson, & Sanders, 1998), and there is growing evidence that this is true for fathers
as well (Buerkel-Rothfuss, Fink, & Buerkel, 1995). Further, mothers talk differently with
daughters than with sons. For example, in infancy, mothers imitate daughters’ vocalizations
more than they do sons’ (Masur, 1987). Even during the first year of life, mothers make more
interpretations (statements concerning the infants’ feelings, needs, or wishes) and engage in
more ‘conversation’ with daughters than with sons, while they are more likely to make more
comments (labeling or naming) to sons and to use words that are intended to draw their sons’
attention (Clearfield & Nelson, 2006). Mothers also use more supportive language with
daughters than with sons during both toddler and school-age years (Leaper et al., 1998), and
tend to be less directive with sons than with daughters once children are school age, perhaps
as a way of encouraging male autonomy. Note that as a whole, these findings show that moth-
ers (and perhaps fathers) engage in more conversational, supportive, and interpretive com-
munication with daughters than with sons from infancy through (at least) elementary school.
Research on gender differences in emotion socialization, while not directly concerned
with language development, has identified language as a primary pathway for the gender
socialization of emotion (Fivush, 1989, 1991; Kuebli, Butler, & Fivush, 1995) and thus
has implications for the gender socialization of language itself. From this area of research
we know that both mothers and fathers use a greater number and variety of emotion
words with daughters than with sons during the preschool years (Fivush, 1989; Fivush,
Brotman, Buckner, & Goodman, 2000; Kuebli & Fivush, 1992). During childhood and
adolescence, mothers discuss emotions differently with daughters and sons, focusing
with daughters on internal feeling states and the resolution of feelings within complex
social interactions, and with sons on the causes and consequences of their actions and on
practical problem-solving skills (Fivush, 1989, 1991; Kuebli et al., 1995).
In both areas of research, differential parent behavior with daughters and sons, including
their verbal behavior, is assumed to reflect parents’ conscious or unconscious conformity
with their society’s gender roles (Clearfield & Nelson, 2006; Golombok & Fivush, 1994;

Downloaded from fla.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016


4 First Language XX(X)

Leaper et al., 1998). For example, in US society females are expected not only to ‘be’ emo-
tional, but to manage the emotional work of the family (Gilligan, 1982; Golombok &
Fivush, 1994; Maccoby, 1990), while males are expected to ‘hold feelings in’ and to be
action-oriented (Levant et al., 2003; Wong, Pituch, & Rochlen, 2006). Teaching daughters
to focus on feeling states and to resolve feelings within complex verbal interactions, and
sons to focus not on internal feeling states but on the causes and consequences of their
actions and on practical problem-solving strategies, reflects not only parents’ own personal
gender beliefs, but their understanding of the gender-relevant behavior that will be expected
from their children later in life. The focus on nuance and complexity in conversations with
daughters may contribute to the early development of complex language skills in a way that
the more straightforward conversations with sons may not. Gender differences in early
language skills may thus reflect the differential reinforcement of daughters and sons for
verbal behaviors that are associated with stereotypical gender roles.
Although these areas of research assume that parent socialization efforts reflect parent
gender-role expectations, neither deals directly with the influence of gender socialization
on mothers’ and fathers’ own verbal behavior. However, mothers’ and fathers’ verbal
behaviors are also products of early gender socialization and are thus likely to be
different from each other. To explore this idea we need to turn to a third literature.

Fathers’ Influence on Language Development


Much of the early research on language development was conducted with mothers and their
children. Mothers were generally more available to researchers than fathers and were expected
to have both more contact with their children and more familiarity with and influence on their
children’s language development than fathers. Cultural shifts and controversies about gender
roles within the family have heightened academic interest in fathering and research on fathers
has grown tremendously over the last 30 years (Goldberg, Tan, & Thorsen, 2009). Research on
fathers’ participation in early conversations with children has grown accordingly. While this
literature has not focused on gender differences in child language development, it has impor-
tant implications for understanding gender effects in early parent–child verbal interaction.
There are many findings of similarities in mothers’ and fathers’ conversations with their
children (Abkarian, Dworkin, & Abkarian, 2003), but there are also areas in which differ-
ences are well documented. For example, there is ample evidence that fathers talk less to
young children than mothers do (Davidson & Snow, 1996; Leaper et al., 1998; McLaughlin,
White, McDevitt, & Raskin, 1983; Rondal, 1980). This has been demonstrated in a variety of
contexts within the home, in father-only interactions with children, and in interactions in
which both parents participate. At least during the preschool years, fathers use fewer pitch
fluctuations, are less sensitive to their children’s attentional focus, are less likely to continue
a child’s topic of conversation, and are less skillful in supporting and scaffolding their chil-
dren’s conversations than mothers (Abkarian et al., 2003; Leaper et al., 1998; Mannle &
Tomasello, 1987; Tenenbaum & Leaper, 1998). They also ask for more clarifications of their
children’s utterances, which is interpreted as reflecting a greater level of breakdown in father–
child than in mother–child conversations (Mannle & Tomasello, 1987; Rowe, Coker, & Pan,
2004). Other researchers have reported that fathers use a more varied vocabulary, more rare
and abstract words, and are likely to ask more ‘w’ questions – ‘who,’ ‘what,’ ‘when,’ ‘where,’
‘why’ – than mothers (Leaper et al., 1998; Masur & Gleason, 1980; McLaughlin et al., 1983;

Downloaded from fla.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016


Lovas 5

Rowe et al., 2004; Walker & Armstrong, 1995). Some research suggests that these differences
diminish by the time children enter school (Davidson & Snow, 1996; Leaper et al., 1998),
perhaps as a result of children’s increasing facility with language. Such findings suggest that
fathers are less ‘fine-tuned’ to their children’s early conversation than mothers and place more
demands on their children for language competence.
The resultant lack of paternal fine-tuning appears to have benefits. Fathers have been
described as acting as a ‘bridge’ for their children, helping them stretch and expand their
verbal skills in order to communicate both with him and with individuals outside the
family sphere (‘the bridge hypothesis,’ see, e.g., Barton & Tomasello, 1994; Gleason,
1975; Mannle & Tomasello, 1987; Rowe et al., 2004). There is indeed research support-
ing the contention that children stretch their abilities when communicating with their
fathers. For example, researchers have found that in some circumstances preschool-age
children (2;0–4;0) produce a more varied vocabulary and use longer and more complex
utterances with fathers than with mothers (Leaper & Gleason, 1996; Masur & Gleason,
1980; Rowe et al., 2004). The bridge phenomenon has thus been interpreted in a positive
light and is assumed to affect both sons and daughters equally.
One explanation for this phenomenon is that fathers spend less time with their children than
do mothers (Mannle & Tomasello, 1987; Rowe et al., 2004). This was true in the 1970s and
1980s when such research was first conducted, and remains true today, despite increased num-
bers of mothers in the workforce (Crompton, Brockmann, & Lyonette, 2005), an increased
public acceptance of the ideal of egalitarian parenting (Milkie, Bianchi, Mattingly, & Robinson,
2002), and increased attention to fathering in academic discourse (Goldberg et al., 2009).
Research shows that while the proportion of time fathers spend with children relative to mothers
has grown – in direct proportion to the increase in mothers’ work outside the home (Jones &
Heermann, 1992) – mothers continue to spend more time with children and take on more
responsibility for day-to-day primary care than fathers (Craig, 2006; Kalenkoski, Ribar, &
Stratton, 2006; Lamb & Oppenheim, 1989; Maume, 2008). Family roles are clearly in flux: the
gap between mothers and fathers in primary hands-on care is decreasing slightly (Sayer, Bianchi,
& Robinson, 2004), but this appears to be driven by changes in a narrow segment of society
rather than by broad changes in family roles across the entire population. Mothers are thus still
likely to engage in more conversations about daily activities, to be more aware of their children’s
vocabulary and language development, and to engage in more complex and varied conversa-
tions with their children than fathers. Further, children engage in more varied and complex
language in situations which are familiar and routinized (Conti-Ramsden & Friel-Patti, 1987),
and are likely to spend more time with their mothers in such activities than with their fathers.
This research implicitly assumes that gender-role socialization impacts the structure of
the family and thus indirectly shapes parent verbal behavior. It is possible that there are
also more direct effects of parent gender socialization on their language with children, but
these have not yet been explored. More importantly, this literature has not explored
possible differences in the verbal behavior of father–daughter and father–son dyads.

Statement of the Problem


A synthesis of these three literatures suggests that parents’ verbal behavior with their children
is affected by their children’s behavior, by their own beliefs about gender roles, and by the
amount of time they spend with their children, whereas children’s verbal behavior is affected

Downloaded from fla.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016


6 First Language XX(X)

by their own biological readiness for verbal communication and by their parents’ verbal
behavior toward them. Parent effects should be stronger than child effects, especially in very
early development when child effects might be driven mostly by small biological differences,
but child effects would be expected to grow over time as parental reinforcement widens these
differences. The major problem with this model is conflict between evidence in support of the
‘bridge hypothesis’ (Barton & Tomasello, 1994; Mannle & Tomasello, 1987), which suggests
that fathers’ demands for verbal clarity and complexity affect both sons and daughters equally,
and evidence that child gender differences in language development are fairly robust across
this time period. Research that carefully attends to the interplay of gender influences in and
among all four types of dyadic relationships (mother–daughter, mother–son, father–daughter,
father–son) would go a long way toward resolving this conflict and establishing a stronger
theoretical framework for understanding gender differences in early language development.

Current Study
The current study explores gender influences on early verbal behavior by studying language
development across time within the context of both mother–toddler and father–toddler
dyads. This approach provides an opportunity to explore child, parent, and dyadic gender
effects on language development as well as the relative importance of child versus parent
influences over time. Data consist of parent reports on two major markers of early language
development (number of words and upper limit mean length utterances or MLUs), collected
when the toddlers were 1;7 and 2;0. This time span brackets the language spurt at the end of
the second year (Bates, O’Connell, & Shore, 1987; Huttenlocher et al., 1991), when the
rapid increase in language development may allow greater visibility of influential variables.

Hypotheses
H1: Because mothers and fathers have been reported to speak more with daughters than
with sons, and both mothers and fathers engage in more emotionally complex conversa-
tion with daughters than with sons, it was expected that parents would report more words
and longer upper limit MLUs for daughters than for sons at both 1;7 and 2;0.
H2: Because mothers are reported to talk more with children than fathers and engage in
more and longer conversations with children about a wider range of day-to-day experi-
ence, mothers were expected to report more words than fathers at both 1;7 and 2;0. As
there is evidence that fathers elicit more complex language from children than mothers
during preschool years, it was expected that longer upper limit MLUs would be reported
by mothers at 1;7, but by fathers at 2;0.
H3: Extrapolating from predictions that daughters would use more words than sons and
mothers would report more words than fathers at both 1;7 and 2;0, it was predicted that
at both 1;7 and 2;0 the most words would be reported in mother–daughter dyads and the
fewest in father–son dyads.
H4: Extrapolating from predictions that at 1;7 daughters would display longer upper
limit MLUs than sons and mothers would report longer upper limit MLUs than fathers,
it was predicted that the 1;7 dyadic pattern for upper limit MLUs would be similar to that
for words: the longest upper limit MLUs would be reported in mother–daughter dyads
and the shortest in father–son dyads. Extrapolating from predictions that at 2;0 daughters

Downloaded from fla.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016


Lovas 7

would display longer upper limit MLUs than sons but that fathers would report longer
upper limit MLUs than mothers, it was predicted that the 2;0 dyadic pattern for upper
limit MLUs would look quite different, with the longest upper limit MLUs reported in
father–daughter dyads and the shortest in mother–son dyads.

Method
Participants
The sample consists of mother–toddler and father–toddler dyads from 113 families (226
adults, 50 girls, 63 boys), solicited through birth records at local hospitals near two uni-
versity towns in California’s central valley.2 Recruitment at both sites (the project started
at one and moved to the other) drew from a range of rural, small town, and urban envi-
ronments. Letters and response cards were sent to English-speaking families with infants
in the target age range.3 Parents were told that the project involved ‘interactive play’
between toddlers and their parents as it related to ‘various aspects of development.’ All
the families recruited from the second site (90 families) were invited back for a second
visit.4 Fifty-seven families (114 adults, 29 girls, 28 boys) returned.
At the time of first data collection, mothers’ ages ranged from 19;0 to 44;0 (M = 32;4.8,
SD = 5;2.4); fathers’ from 21;0 to 50;0 (M = 34;6, SD = 5;10.8); and children’s from 1;6 to
1;8.2 (M = 1;7.2, SD = 0;0.4). At the second visit, children ranged in age from 1;11.9 to
2;1.9 (M = 2;0.5, SD = 0;0.4). One-hundred-and-eight couples (96%) were married; four
were cohabiting, and one couple was separated. Forty percent of the families had one child;
60% had older children (range of siblings = 0–4, M = 0.96, SD = 1.01). A rough estimate of
ethnicity indicated the parents were 89% Caucasian, 6% Hispanic, 4% Asian American,
and 1% African American;5 14% of the families were multi-ethnic. The families were pre-
dominantly middle-class, with 22% of annual family incomes below $30,000, 34% between
$30,000 and $50,000, and 44% over $50,000. The parents were well educated: 99% had
finished high school, over 50% had finished college, and about a quarter (28% of mothers
and 24% of fathers) had completed at least some graduate work. Fifty-seven percent of
fathers and 46% of mothers scored as semi-professional or above on an adapted version of
the Hollingshead–Redlich Occupational Scale (1958).6 Fathers’ occupations were nor-
mally distributed. Although there were no educational differences between mothers and
fathers, the distribution of mothers’ occupations was bimodal, with 35% in a lower level
that indicated they were home with their children full-time. Of the 65% who were working,
many had part-time jobs; daycare for these families averaged about 25 hours per week.

Procedure
Data were collected during visits to developmental laboratories when observations of
mother–toddler and father–toddler dyads (for purposes of the larger project) were sched-
uled. The mother, father, and toddler came to the lab together and met with the experi-
menter in an office. Consent forms were collected and demographic information forms
were completed. Each parent filled out a set of questionnaires while the other parent was
observed with the child. Procedures for the 2;0 follow-up were identical, except that
demographic information was not collected.

Downloaded from fla.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016


8 First Language XX(X)

Measure
Parent reports of language development were obtained with the MacArthur
Communicative Development Inventory: Toddlers (the ‘CDI;’ Fenson et al., 1991b).
Two variables were calculated for the current study. Words is a marker of vocabulary
production and consists of a simple word count. The CDI provides a list of 680 words in
22 categories commonly used by children (e.g., ‘animals,’ ‘food and drink,’ ‘words about
time’). Parents were asked to note which of these words they had heard their child say,
regardless of whether the child’s pronunciation was correct or whether the child used
another form of the word (such as ‘sketti’ for ‘spaghetti’). About a third of the sample
neglected to fill out the last two usage categories (‘helping verbs’ and ‘connecting
words’). Independent t-tests comparing truncated total scores (categories 1–20 only)
found no differences between the parents who completed the last two questions and those
who did not (mothers, t(111) = 0.551, p = .55; fathers, t(111) = 1.647, p = .102). The
truncated totals were therefore used as the word count scores.
The second CDI variable, mean length utterances (MLUs), is a marker both of
sentence length and of grammatical complexity. Parents were asked to provide ‘the
three longest sentences’ they had heard their child use. All three sentences were then
coded for morphemes – meaningful units of language. ‘Mrs Brown,’ for example,
counts as one morpheme because it is a name. Likewise, words that are repeated are
only counted once, since they do not add new meaning to the sentence. However, the
added ‘s’ for a plural or ‘ed’ for past tense add an extra morpheme to words since they
add meaning. ‘Dogs’ counts as two morphemes, for example, as would ‘eated.’ After
the separate sentences were scored, the scores were averaged to produce a final score.
Since research shows that CDI MLUs are typically higher than averages computed on
a larger sample of sentences from actual transcripts (Dale, 1991), the designation
upper limit MLUs will be used for the MLUs in the current paper.

Reliability and validity.  The MacArthur CDI has been widely used and its reliability and
convergent validity, both with other methods of assessing language development and
with measures of intelligence, has been well established (for details, see Bornstein &
Haynes, 1998; Dale, 1991; Fenson et al., 2000; Pan, Rowe, Spier, & Tamis-Lemonda,
2004; Ring & Fenson, 2000).

Stability.  Bornstein and colleagues have assessed the stability of CDI scores in several
studies. In one they found moderate to high stability for both father and mother reports
between 1;0 and 1;8 (Bornstein, Putnick, & De Houwer, 2006). Although they did not
analyze for differences among reporters, fathers’ stability correlations were slightly
higher than mothers’. This study did not analyze for child gender differences. In another
report on four longitudinal studies across various age spans, Bornstein et al., (2004),
found moderate to high stability for mother reports throughout the period between 1;1
and 6;10, and no child gender differences. These studies generally included CDI MLUs
in a composite score; the single finding that calculated specific stability for MLUs found
low but significant stability for mother reports of MLUs between 1;8 and 4;0 (Bornstein
et al., 2004).

Downloaded from fla.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016


Lovas 9

Coding and interrater reliability.  The MLU coding information provided with the MacArthur
CDI Technical Manual (Fenson et al., 1991a) was not sufficient for coding. Two additional
sources (Fazio & Johnson, 1998; Miller, 1981) provided enough clarification to proceed
with the development of an MLU training manual. Two undergraduate research assistants
were trained by the author to code the MLUs. Once they established reliability with the
author they began coding the MLUs separately in order to establish reliability with each
other. Thirty-six percent of the sample was coded by both coders, who met periodically to
compare their scores and resolve discrepancies. Interclass correlation coefficients (two-
way random, absolute agreement criteria) on the reliability sample ranged from .97 to 1.00.

Analyses
To check the data for stability, the longitudinal data were ranked and correlated across time
using Spearman’s rho. Contrasts were analyzed by converting the correlations to Fisher’s
Zs and calculating z-scores on differences between the Fisher’s Zs (see Howell, 1992).
Repeated measures MANOVAS were run on the 1;7 (entire sample) and 2;0 (longitudi-
nal subsample) data, with child gender as a between-subjects variable, parent gender as a
within-subjects variable, and words and upper limit MLUs as dependent variables. Dyadic
contrasts were explored with paired (MD/FD, MS/FS) and independent (MD/MS, MD/FS,
MS/FD, FD/FS) t-tests. Effect size analyses (using Cohen’s d) supplemented both the
MANOVAs and the dyadic t-tests. Significance levels for all tests involving more than two
groups and for all groups of correlations are reported with appropriate Bonferroni adjust-
ments.7 Because scores for words (range = 2.5–598.5) and upper limit MLUs (range = 1.0–
9.25) have a different scale, figures comparing words and upper limit MLUs use z-scores.
Note that this study focuses on the dyad as a unit of analysis. Although main effects and
interactions are noted, the analysis of dyadic scores allows for direct comparisons between and
among the dyads and provides a basis for comparing patterns of scores across the dyads (within
each time period separately). The use of effect sizes is crucial because effect sizes (Cohen’s d;
Cohen, 1988) are generally more sensitive to small but potentially important group differences
than significance testing.8 In the current study, the model assumes that even very small child
gender differences could be meaningful in the context of bidirectional feedback mechanisms
– that is, they may impact parent behavior and be amplified by reinforcement over time.
Note also that the focus of these analyses (with the exception of the stability check)
was not on change over time within groups but on differences among groups within time.
Although figures provide a visual representation of dyadic means and correlations at
both 1;7 and 2;0 and changes in the dyadic pattern of means and correlations between 1;7
and 2;0 will be discussed, reported contrasts refer to differences among groups within
each time period rather than differences over time.9

Results
Means and standard deviations can be found in Table 1. The expected vocabulary spurt
was reflected in vocabulary increases between 1;7 and 2;0 of 349.8% for daughters and
317.4% for sons. Upper limit MLUs also increased during this period by 207.9% for
daughters and 184.7% for sons.

Downloaded from fla.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016


10 First Language XX(X)

Table 1.  Means and standard deviations for words and MLUs

Words MLUs

N M SD M SD

At 1;7 – entire sample


  Child gender
   Daughters 100   89.82   82.53 2.15 1.07
   Sons 126   83.63   88.91 1.90 0.90
  Parent gender
   Mothers 113   95.03   88.37 2.17 0.97
   Fathers 113   77.71   83.08 1.86 0.97
  Dyads
   Mother–daughter   50   97.58   84.06 2.35 1.05
   Mother–son   63   93.00   92.27 2.03 0.90
   Father–daughter   50   82.06   81.08 1.96 1.07
   Father–son   63   74.25   85.12 1.78 0.89
At 2;0 – longitudinal sample only
  Child gender
   Daughters   58 314.22 142.40 4.47 2.32
   Sons   56 265.46 174.60 3.51 1.87
  Parent gender
   Mothers   57 301.09 145.05 4.05 1.98
   Fathers   57 279.46 174.68 3.94 2.34
  Dyads
   Mother–daughter   29 340.62 131.34 4.29 2.13
   Mother–son   28 260.14 149.44 3.81 1.80
   Father–daughter   29 287.83 150.29 4.65 2.51
   Father–son   28 270.79 199.28 3.20 1.91

Stability Over Time


Stability was assessed on the subsample of the 57 toddlers and 114 mothers and fathers
who provided data at both 1;7 and 2;0. As can be seen in Table 2, parents’ reports
of words were highly stable across time for all groups (rs = .57, p < .001 to rs = .86,
p < .001). This was consistent with reports by Bornstein and colleagues (Bornstein et al.,
2004, 2006). Stability correlations for daughters and mothers were significantly lower
than those for sons and fathers (z = –2.64, p < .01; z = –.253, p < .05), which is consistent
with the Bornstein et al. (2006) data, at least for mothers and fathers. Parents’ reports of
upper limit MLUs were less stable than those for words (consistent with Bornstein et al.,
2004), but were significant for all groups except father–child dyads (rs = .27, p < .05 to
rs = .59, p < .001). The stability of upper limit MLUs was significantly lower for fathers
than for mothers (z = 2.76, p < .01).
Dyadic contrasts among stability correlations are presented in Table 2, but the con-
trasts and patterns of correlations can be seen more easily in Figure 1. There were

Downloaded from fla.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016


Table 2.  Stability correlations and contrasts for ranked data across time (longitudinal sample only)
Lovas

Words MLUs

N rs p z p rs p z p

Child gender
   Daughters 58 0.70 *** 0.41 ***
   Sons 56 0.84 *** 0.40 ***
   Contrast –2.64 ** 0.09 NS
Parent gender
   Mothers 57 0.69 *** 0.57 ***
   Fathers 57 0.83 *** 0.27 *
   Contrast –2.53 * 2.76 **
Dyads
   Mother–daughter 29 0.57 *** 0.59 ***
   Mother–son 28 0.81 *** 0.55 **
   Father–daughter 29 0.81 *** 0.33 NS
   Father–son 28 0.86 *** 0.22 NS
Dyadic contrasts
   Mother–daughter/Father–son 57 –3.36 *** 2.36 *
   Mother–daughter/Father–daughter 58 –2.52 * 1.76 t
   Mother–son/Father–son 56 –0.86 2.03 *
   Mother–daughter/Mother–son 57 –2.49 * 0.31
   Mother–son/Father–daughter 57 0.00 1.43

Downloaded from fla.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016


   Father–daughter/Father–son 57   –0.86   0.62 

Note: Correlations calculated with Spearman’s rho, z-scores calculated with Fisher’s Zs.
Significance levels for two correlations reported at Bonferroni equivalents for: .01*** (.005), .05** (.025), or at unadjusted .05*.
Significance levels for four correlations reported at Bonferroni equivalents for: .01*** (.0025), .05** (.0125), or at unadjusted .05*.
Significance levels for six contrasts reported at Bonferroni equivalent of: .001*** (.0016), .01** (.0083), or at unadjusted .05*.
11
12 First Language XX(X)

significant contrasts between MD and FS correlations for both words (z = –3.36, p < .001)
and upper limit MLUs (z = 2.36, p < .05), with father–child dyads showing higher stability
than mother–child dyads for words, and lower stability than mother–child dyads for upper
limit MLUs. In fact, stability for father–child dyads failed to reach significance (rs = .22,
NS; rs = .33, NS).

Multivariate Effects
As can be seen in Table 3, there was a significant multivariate main effect for parent
gender at 1;7, F(2, 110) = 10.98, p < .001, and a significant interaction between child and
parent gender at 2;0, F(2, 54) = 5.77, p < .01. These findings support the expectation that
at least by 2;0, both children and parents were contributing to gender differences in
reported language development.

Univariate Main Effects and Effect Size Contrasts


Anticipated child gender effects for words and upper limit MLUs were not generally
supported by the univariate analyses. However, as can be seen in Table 3, the effect
size analyses extended and elaborated upon these findings. At 1;7, there was a small
effect size contrast for upper limit MLUs favoring daughters (d = .26). At 2;0, there
was a small effect size contrast for words (d = .30), and both a univariate trend and a
small effect size contrast for upper limit MLUs, F(1, 55) = 3.60, p = .07 (d = .45), all
favoring daughters.

Figure 1.  Dyadic stability correlations and contrasts across time

Downloaded from fla.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016


Lovas

Table 3.  Main effects and interactions: Multiple analyses of variance and effect sizes

Univariate

Multivariate Words MLUs


a
d.f. F p d.f. F p d ES F p d ESa

At 1;7 – entire sample


   Between subjects
    Child gender    2 1.28    1 0.16 0.07 2.45 0.26 S
  Within subjects
    Parent gender    2 10.98 ***    1 12.59 *** 0.20 S 12.94 *** 0.32 S
    Child gender × Parent gender    2 0.40    1 0.11 0.59
   Error 110 111
At 2;0 – longitudinal sample only
   Between subjects
    Child gender    2 1.80    1 1.59 0.30 S 3.60 t(.07) 0.45 S
  Within subjects
    Parent gender    2 0.76    1 1.55 0.14 0.26 0.05
    Child gender × Parent gender    2 5.77 **    1 3.51 t(.07) 4.18 *
   Error   54       55

Downloaded from fla.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016


a
Effect sizes (ES) calculated with Cohen’s d (M1–M2/SD12). Evaluations based on Cohen’s criteria: S > .10, M > .50, L > .80.
***p < .001, **p < .01, *p < .05, t < .10.
13
14 First Language XX(X)

Predictions of parent gender effects favoring mothers at 1;7 were confirmed for both
words, F(1, 111) = 12.59, p < .001 (d = .20, small effect size) and upper limit MLUs, F(1,
111) = 12.94, p < .001 (d = .32, small effect size). Neither prediction for parent effects at
2;0 (that mothers would report more words than fathers but that fathers would report
longer upper limit MLUs ) were supported.

Univariate Interactions, Dyadic t-tests, and Effect Size Contrasts


Univariate interactions between child and parent gender were not significant at 1;7, but
Bonferroni adjusted t-tests and effect size analyses of dyadic contrasts extended and
elaborated upon these findings. As can be seen in Table 1, reported words and upper limit
MLUs for dyads at 1;7 displayed a descending pattern of scores from mother–daughter
(MD) to mother–son (MS) to father–daughter (FD) to father–son (FS) dyads. This con-
firmed the expectation that at 1;7, mother–daughter dyads would display the highest
scores and father–son dyads the lowest for both words and upper limit MLUs. The con-
trast for words reached small effect size (d = .27); the contrast for upper limit MLUs was
significant, t(111) = 3.15, p < .01 (d = .57, medium effect size). These and other, smaller
contrasts are presented in Table 4; z-scores and effect size contrasts for words and upper
limit MLUs at 1;7 can be seen in Figure 2.
At 2;0, univariate interactions were of borderline significance for words, F(1, 55) =
3.51, p = .07, but clearly significant for upper limit MLUs, F(1, 55) = 4.18, p < .01. The
pattern of dyadic scores for words was not entirely consistent with expectations. As can
be seen in Table 1, the most words were reported in mother–daughter dyads (as expected)
but the fewest in mother–son rather than father–son dyads (MD/MS contrast, t(55) =
2.16, p < .05, d = .55, medium effect size). The longest upper limit MLUs were reported
in father–daughter dyads (also as expected), but the shortest in father–son rather than
mother–son dyads (FD/FS contrast, t(55) = 2.44, p < .05, d = .62, medium effect size).
These and other, smaller contrasts are presented in Table 4; z-scores and effect size
contrasts for words and upper limit MLUs at 2;0 can be seen in Figure 3.

Discussion
In general, child gender differences in reported words and upper limit MLUs increased
over time while parent gender reporting differences decreased. For children, this result is
consistent with the literatures on gender differences in neurological maturity and organi-
zation, the gender socialization of language, and gender differences in emotion socializa-
tion. Whether child gender differences are reinforced by parent behavior or are first
created and then reinforced, such differences should be evident by 1;7 and should con-
tinue to increase over time. Parents’ beliefs and expectations about gender roles, their
own gender-differentiated verbal behavior, and their children’s verbal attention and
behavior are all expected to influence their verbal behavior toward sons and daughters,
creating feedback loops of mutual influence that create increasingly larger child gender
differences over time.
The decrease in parent gender reporting differences over time was not predicted.
Mothers had been expected to report more words than fathers at both 1;7 and 2;0, and a

Downloaded from fla.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016


Lovas

Table 4.  Dyadic contrasts: paired and independent t-tests and effect sizes

Words MLUs

d.f. t p d ESa t p d ESa

At 1;7 – entire sample


   Mother–daughter/Father–son 111 1.46 0.27 S 3.15 ** 0.57 M
   Mother–daughter/Father–daughter 49 2.59 * 0.19 2.58 * 0.36 S
   Mother–son/Father–son 62 2.62 * 0.21 S 2.41 * 0.28 S
   Mother–daughter/Mother–son 111 0.27 0.05 1.75 0.33 S
   Mother–son/Father–daughter 111 0.66 0.13 0.38 0.07
   Father–daughter/Father–son 111 0.49 0.09 0.99 0.19
At 2;0 – longitudinal sample only
   Mother–daughter/Father–son 55 1.57 0.41 S 2.02 * 0.52 M
   Mother–daughter/Father–daughter 28 2.09 * 0.37 S -1.00 -0.16
   Mother–son/Father–son 27 -0.48 -0.06 2.00 0.32 S
   Mother–daughter/Mother–son 55 2.16 * 0.55 M 0.91 0.24 S
   Mother–son/Father–daughter 55 -0.70 -0.19 -1.45 -0.40 S
   Father-Daughter/Father-Son 55 0.37   0.10   2.44 * 0.62 M
Note: Paired contrasts (MD/FD, MS/FS) analyzed with paired t-tests; other contrasts analyzed with independent t-tests. For longitudinal analysis, scores averaged across

Downloaded from fla.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016


time before computing paired and independent t-tests.
a
Effect sizes (ES) calculated with Cohen’s d (M1–M2/SD12). Evaluations based on Cohen’s criteria: S > .10, M > .50, L > .80.
Significance levels for four correlations reported at Bonferroni equivalents for: .01*** (.0025), .05** (.0125), or at unadjusted .05*.
Significance levels for six contrasts reported at Bonferroni equivalent of: .001*** (.0016), .01** (.0083), or at unadjusted .05*.
15
16 First Language XX(X)

Figure 2.  Dyadic z-scores and effect size contrasts at 1;7

Figure 3.  Dyadic z-scores and effect size contrasts at 2;0

parent gender difference for upper limit MLUs was expected to reverse over time, with
longer upper limit MLUs reported by mothers at 1;7, but by fathers at 2;0. While the
expected parent gender differences in both words and upper limit MLUs were evident
at 1;7, they had disappeared by 2;0.

Downloaded from fla.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016


Lovas 17

As can be seen in Figure 3, however, the similarity in mother and father scores at 2;0
both reflects and obscures important patterns of dyadic difference. Mothers did, as
expected, report more words than fathers at 2;0, but only for daughters. The very large
MD/MS difference in words at 2;0 cancels out in the main effects, as does the much
smaller non-significant FD/FS difference. Likewise, fathers did, as expected, report lon-
ger upper limit MLUs than mothers at 2;0, but again, only for daughters. Here, too, the
very large FD/FS difference in upper limit MLUs at 2;0 cancels out in the main effects,
as does the much smaller MD/MS difference.
What is more surprising at 2;0 is the apparent importance of child gender as an influ-
ence on words in the context of mother–child dyads, and on upper limit MLUs in the
context of father–child dyads. In the case of words, a closer look at the data suggest that
the strongest shift over time was that mother–daughter dyads pulled ahead of the other
dyads. Although mother–son dyads displayed the smallest increase in words, there was
actually little difference among mother–son, father–daughter, and father–son dyads in
reported words at 2;0. In the case of upper limit MLUs, the strongest shift over time was
that father–son dyads had dropped behind the others in complex language use. The lon-
gest upper limit MLUs at 2;0 were indeed reported in father–daughter dyads, but the
difference between father–daughter and mother–daughter dyads was not significant. The
more telling finding is that for both words and upper limit MLUs, parent–daughter
reports were significantly higher than parent–son reports.
What this pattern suggests is that the 2;0 results may have been driven largely by child
gender and can thus be interpreted as providing fairly strong evidence of bidirectional
influences. By 2;0, mothers and fathers may still be using language differently with sons
and daughters, but sons and daughters may be increasingly influencing the communica-
tion they have with their parents. Daughters’ greater facility with and interest in language,
even at this young age, may be just as important as parents’ conscious or unconscious
socialization efforts – or even their familiarity with their children’s verbal abilities – in
driving the increasing differentiation between language use with daughters and sons. The
2;0 findings thus conform to the model of bidirectional influence described earlier.
Further, these results provide evidence that it may in fact be daughters who are driving
much of the change in complex language behavior with fathers during the time period
studied here. A look back at the stability analyses, which found very low stability in
father–child scores for upper limit MLUs, supports this possibility. Low stability
scores for father–child dyads suggests that fathers’ reports were affected very strongly
by gender differences in their children’s levels of sentence complexity, especially at
2;0. While this seems counterintuitive, research on emotional availability (EA) has
found higher stability in EA scores for father–child than mother–child dyads, even
when actual EA scores were lower for father–child than mother–child dyads (Lovas,
2005). The interpretation of these findings was that the high level of stability in fathers’
EA scores reflected their lower levels of responsiveness. In other words, fathers were
more rigid than mothers and their behavior was thus more likely to be based on a ‘cus-
tomary’ approach to interaction with their child rather than on the signals and expecta-
tions of the child at any particular time. This translates well to the current situation,
where the low level of stability for upper limit MLUs in father–child dyads may well
indicate that fathers were increasingly influenced by their children’s levels of commu-
nicative interest and linguistic complexity by 2;0.

Downloaded from fla.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016


18 First Language XX(X)

These findings also help explain anomalies in research on the bridge hypothesis. For
example, Davidson and Snow (1996) reported that in a sample of 5-year-olds and their
parents, who were observed in several contexts within the home (dyadic and triadic),
mother–child dyads consistently displayed not only more language but more complex
language than father–child dyads. But their sample consisted of only 12 child partici-
pants, eight of whom were sons. The lower scores for father–child dyads may therefore
have reflected exactly what we see here in a much larger sample of children: that father–
son dyads substantially trail behind mother–child and father–daughter dyads in complex
language use.
Other research finding that children use more complex language with fathers during
the preschool years (Leaper & Gleason, 1996; Masur & Gleason, 1980) may be under-
representing the size of this effect for father–daughter dyads and overrepresenting it for
father–son dyads. For example, Masur and Gleason (1980) and Leaper and Gleason
(1996) both report a child effect favoring daughters and a parent effect favoring fathers,
but did not report whether father–daughter and father–son dyads differed from each
other. Their samples were also quite small (14 children and 24 children, respectively).
The size of the current sample gives the present findings more power and generalizabil-
ity, and argues for a closer look at this phenomenon.
It is quite possible, as Leaper and Gleason (1996) argue, that the outcomes seen here
are driven not so much by language capacity but by the kinds of activities in which these
dyads most often engage (see also, Lewis & Gregory, 1987). We know that parent–son
dyads, and especially father–son dyads, are more likely than parent–daughter dyads to
engage in active, less narrative play (Leaper & Gleason, 1996; O’Brien & Nagle, 1987),
and this certainly affects both the time and motivation available for extended conversa-
tion. This does not, however, lessen the impact of these differences on expectations about
and facility in communication that are being constructed in these early years.

Theoretical Implications
These findings have implications not only for gender differences in language use but for
gender development more generally. Regardless of differences in the contexts for lan-
guage use between parent–daughter and parent–son dyads, or whether the differences are
being driven by increasing parent differentiation between sons and daughters or by
increasing differences in the language ability of daughters and sons, the emerging picture
is one in which daughters are being given ample opportunity to practice verbal exchange
with both parents, while sons are learning less about verbal exchange with both mothers
and fathers, but especially with fathers. Sons may therefore be learning very early that
talking with complexity is not something that is done or is encouraged, especially with
other males and perhaps not with anyone.
This message is consistent with many aspects of later male gender-role functioning,
especially when taken in conjunction with well-established findings that conversations
with daughters are generally richer both in emotion words and complex conversations
about emotions than those with sons (Fivush, 1989, 1991; Fivush et al., 2000; Kuebli
et al., 1995; Kuebli & Fivush, 1992). For example, many researchers and clinicians point
to restricted emotionality as one of the central themes in masculine ideology and thus in

Downloaded from fla.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016


Lovas 19

male identity (Jansz, 2000; Kilmartin, 2007; O’Neil, Good, & Holmes, 1995). Restricted
emotionality, in turn, has been linked to depression, anxiety, and other mental health
issues (Pollack, 2005; Shepard, 2002), to difficulty with emotional communication
(Hampson, van Anders, & Mullin, 2006; Wong et al., 2006), and even to mild to moder-
ate levels of alexithymia (Levant et al., 2003, 2006). More specifically, restricted
emotionality and male ‘inexpressiveness’ are considered to be major barriers to commu-
nication in intimate relationships (Dosser, Balswick, & Halverson, 1986; Gray, 1992;
Tannen, 1990) and ultimately to the quality of intimate relationships (Dempsey, 2002;
Levant, 1996).
Although many researchers focus on the complementary influences of fathers and
mothers, noting that fathers offer different skills, resources, and experiences than moth-
ers and fulfill different functions with regard to childrearing (Greenspan, 1988; Lamb,
1981; Paquette, Bolte, Turcotte, Dubeau, & Bouchard, 2000; Volling & Belsky, 1992)
– an interpretation also present in literature on the bridge hypothesis (Barton &
Tomasello, 1994; Leaper & Gleason, 1996; Mannle & Tomasello, 1987; Masur &
Gleason, 1980) – it nevertheless remains that the ability to engage in complex commu-
nication is central to the creation of shared experience and intimacy. Seen from this
perspective, the very low reports of upper limit MLUs in father–son dyads is striking.
Whether daughters’ complex language with fathers is driven by fathers’ demands for
complex language or by daughters’ increasing verbal ability, it is clear that the same
factors are not at work in father–son dyads. Considering the fact that fathers provide the
first and often the most accessible model for male gender identity, this finding that
fathers and sons are engaging in less complex conversation at this young age than other
parent–child dyads provides a new vantage point from which to view the trajectory of
gender socialization, especially with regard to the emotional range and complexity of
intimate conversation.

Strengths and Limitations


The major strength of the study is the large sample size which makes the exploration of
parent, child, and dyadic gender effects possible. Further, the use of effect sizes provides
an opportunity to capture small but meaningful differences and to explore the consis-
tency in patterns of difference that emerge from these analyses.
There are several limitations to this study. One major limitation is that the sample is
largely white and middle-class, and provides no information on dyadic parent–child ver-
bal behavior in other populations. An important area for future exploration would be to
examine ethnic variations in these patterns. A more important limitation is that it relies
upon parent reports of child language use rather than upon direct observation. Although
Fenson et al. (1991a, 2000) report good reliability for the CDI, studies that explored
multiple reports of child comprehension have found great variability in the extent to
which parents agreed upon their child’s comprehension of words and sentences (Bornstein
et al., 2006; De Houwer, Bornstein, & Leach, 2005). This is to be expected, since moth-
ers and fathers spend different amounts and kinds of time with their children, and is in
fact relevant to the current study. Another issue is that this study does not include direct
measures of parent language use. Parental verbal behavior is therefore inferred by the

Downloaded from fla.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016


20 First Language XX(X)

assumption that language reports reflect the history of the language use with their
children. Analysis of the actual content of dyadic conversations would extend both the
comprehensiveness and the usefulness of the current analyses.

Conclusion
This study demonstrates differences in the amount and complexity of language use
among mother–daughter, mother–son, father–daughter, and father–son dyads. It pro-
vides evidence of bidirectional influences, especially by 2;0, with both child and par-
ent gender differences in language interactivity contributing to high parent–daughter
scores and low father–son scores, especially in upper limit MLUs. These findings cast
doubt on the generalizability of the bridge effect to father–son dyads and support a
theoretical framework for understanding gender influences in early language develop-
ment. Further exploration of this issue offers the potential to expand our understanding
of gender differences in language development and to help explain later gender
differences in communicative behavior.

Acknowledgments
My thanks to Yvonne Emmons, Susan W. Goodwyn, and Linda P. Acredolo, who granted me
access to their data; to Linda P. Acredolo, Carol J. Rodning, and Leslie R. Brody, who provided
advice, support, and help with various aspects of this project; and to Lisa Krause and Adriana
Morgenstern, the research assistants who coded the MLUs. Sincere thanks as well to the parents
and children who contributed to this research. Funding for this project was provided by a Jean
Humphrey Block Dissertation Grant from the Henry A. Murray Research Center at Radcliffe
College for the period 1 July 2000–30 June 2001; a Graduate Research Award from the Consortium
for Women and Research at UC Davis for the period 1 July 1999–30 June 2001; and a Dissertation
Year Fellowship from the University of California for the period 1 July 2000–30 June 2001.

Notes
1 See evidence of gender differences in brain lateralization and organization in older children in
Burman et al. (2008) and Hanlon et al. (1999), and in young adults in Halari et al. (2006).
2 The language development and demographic data used in this study were gathered by Emmons,
Goodwyn, and Acredolo (1993). The larger project was designed to provide data for exploring
multiple aspects of development in the context of parent–child interactions.
3 Language development was a primary interest for Emmons et al. (1993). They did not have
access to translation services and thus limited the sample to English-speaking families only.
4 This project shifted from one university to another as the student researcher moved from
undergraduate to graduate school. Including site 1 participants in the follow-up was not
practical because of travel time and cost constraints.
5 The restriction to English-speaking families noted in note 3, above, limited the number of
Spanish-speaking families in the sample relative to the demographics of the recruitment areas.
The underrepresentation of Asian American and African American families is an accurate
reflection of recruitment area demographics.

Downloaded from fla.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016


Lovas 21

6 Occupation was requested in a short answer format. The seven-point Hollingshead–Redlich


Scale (1958) was appropriate for coding such data, but did not include the occupations
housewife, homemaker, or mother, or such current technological occupations as data entry or
systems analysis. It was therefore adapted for use in this study by assigning missing occupations
to categories based on logical consistency with the occupations already listed. The occupations
of all 226 parents were then coded independently by two undergraduate research assistants.
Discrepancies were resolved by discussion and consensus between the two raters or with the
author. Interclass correlation coefficients (mixed random effects, absolute agreement criteria)
were: mothers .98, F(113, 113) = 86.48, p < .001; fathers .99, F(111, 111) = 80.35, p < .01.
7 Also note that since the MD/MS and FD/FS have fewer N than the other dyads and require
paired rather than independent t-tests, the analyses for these contrasts have more power than
those for the other comparisons. Effect sizes and significance for these contrasts may therefore
be larger than those for contrasts that are larger in absolute values.
8 Effect sizes (d) provide a standardized measure (standard deviation) of group location relative
to joint variance, an estimate of divergence between groups, and a sense of how meaningful
the difference is in terms of the average person’s ability to perceive it. Cohen defined .20 as
a small effect size (15% divergence, not readily observable to the untrained eye), .50 as a
moderate effect size (33% divergence, readily observable to the untrained eye), and .80 as a
large effect size (47% divergence). Effect size can also be used to provide an estimate of where
the average person in one group falls relative to the other group’s distribution.
9 Also note that the 1;7 results in the text, tables, and figures are based on data from the entire
sample of 113 families while the 2;0 results are based on data from the longitudinal subsample
of 57 families only. Preliminary analysis indicated that the only significant difference between
the two subsamples at 1;7 was that mothers in the nonlogitudinal sample reported Sharter MLUs
than mothers in the longitudinal sample. The visual comparison of 1;7 (entire sample) and 2;0
(longitudinal sample only) data in the figures therefore provides a fairly representative comparison
of change over time, but slightly overrepresents changes in mothers’ reported MLUs.

References
Abkarian, G. G., Dworkin, J. P., & Abkarian, A. K. (2003). Father’s speech to their children:
Perfect pitch or tin ear? Fathering, 1, 27–50.
Barton, M. E., & Tomasello, M. (1994). The rest of the family: The role of fathers and siblings in
early language development. In C. Gallaway & B. J. Richards (Eds.), Input and interaction in
language acquisition (pp. 109–134). New York: Cambridge University Press.
Bates, E., O’Connell, B., & Shore, C. (1987). Language and communication in infancy. In
Handbook of infant development (2nd ed., pp. 149–203). New York: John Wiley & Sons.
Berry, J. W., Poortinga, Y. H., Segall, M. H., & Dasen, P. R. (1992). Cross-cultural psychology:
Research and applications. New York: Cambridge University Press.
Bornstein, M. H., Hahn, C.-S., & Haynes, O. M. (2004). Specific and general language performance
across early childhood: Stability and gender considerations. First Language, 24, 267–304.
Bornstein, M. H., & Haynes, O. M. (1998). Vocabulary competence in early childhood:
Measurement, latent construct, and predictive validity. Child Development, 69, 654–671.
Bornstein, M. H., Putnick, D. L., & De Houwer, A. (2006). Child vocabulary across the second
year: Stability and continuity for reporter comparisons and a cumulative score. First Language,
26, 299–316.

Downloaded from fla.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016


22 First Language XX(X)

Breedlove, S. M. (1994). Sexual differentiation of the human nervous system. Annual Review of
Psychology, 45, 389–418.
Buerkel-Rothfuss, N. L., Fink, D. S., & Buerkel, R. A. (1995). Communication in the father–child dyad:
The intergenerational transmission process. In T. J. Socha & G. H. Stamp (Eds.), Parents, chil-
dren, and communication: Frontiers of theory and research (pp. 63–85). Mahwah, NJ: Lawrence
Erlbaum.
Burman, D. D., Bitan, T., & Booth, J. R. (2008). Sex differences in neural processing of language
among children. Neuropsychologia, 46, 1349–1362.
Clearfield, M. W., & Nelson, N. M. (2006). Sex differences in mothers’ speech and play behavior
with 6-, 9-, and 14-month-old infants. Sex Roles, 54, 127–138.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ:
Lawrence Erlbaum.
Conti-Ramsden, G., & Friel-Patti, S. (1987). Situational variability in mother–child conversations.
In K. E. Nelson & A. van Kleeck (Eds.), Children’s language (Vol. 6, pp. 43–63). Hillsdale, NJ:
Lawrence Erlbaum.
Craig, L. (2006). Does father care mean fathers share? A comparison of how mothers and fathers
in intact families spend time with children. Gender & Society, 20, 259–281.
Crompton, R., Brockmann, M., & Lyonette, C. (2005). Attitudes, women’s employment, and the
domestic division of labour: A cross-national analysis in two waves. Work, Employment, and
Society, 19, 213–233.
Dale, P. S. (1991). The validity of a parent report measure on vocabulary and syntax at 24 months.
Journal of Speech & Hearing Research, 34, 565–571.
Davidson, R. G., & Snow, C. E. (1996). Five-year-olds’ interactions with fathers versus mothers.
First Language, 16, 223–242.
De Frias, C. M., Milsson, L.-G., & Herlitz, A. (2006). Sex differences in cognition are stable over a
10 year period in adulthood and old age. Aging, Neuropsychology, and Cognition, 13, 574–587.
De Houwer, A., Bornstein, M. H., & Leach, D. B. (2005). Assessing early communicative ability: A
cross-reporter cumulative score for the MacArthur CDI. Journal of Child Language, 32, 735–758.
Dempsey, K. (2002). Who gets the best deal from marriage: Women or men? Journal of Sociology,
38, 91–110.
Donelson, E., & Gullahorn, J. E. (1977). Sex differences in the growth and use of language. In
Women: A psychological perspective (pp. 79–94). New York: Wiley.
Dosser, D. A., Balswick, J. O., & Halverson, C. F., Jr. (1986). Male inexpressiveness and relation-
ships. Journal of Social and Personal Relationships, 3, 241–258.
Emmons, Y., Goodwyn, S. W., & Acredolo, L. P. (1993). The father project [raw data]. University
of California, Davis, Department of Psychology, Child Development Lab.
Eriksson, M., & Berglund, E. (1998). Communicative development in Swedish children 8–16
months old: The Swedish Early Communicative Development Inventory – words and gestures.
Reports from the Department of Psychology, U. Stockholm (844), pp. 1–35.
Fazio, B., & Johnson, J. J. (1998). LSA/CAI language analysis program [on-line instructional
program for calculating MLUs]. Bloomington, IN: Indiana University.
Fenson, L., Dale, P. S., Reznick, J. S., Thal, D., Bates, E., Hartung, J. P., et al. (1991a). Technical
manual for the MacArthur Communicative Development Inventories [coding manual]. San
Diego, CA: San Diego State University.

Downloaded from fla.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016


Lovas 23

Fenson, L., Dale, P. S., Reznick, J. S., Thal, D., Reilly, J. S., & Bates, E. (1991b). The MacArthur
Communicative Development Inventory: Toddlers [manual]. San Diego, CA: San Diego State
University.
Fenson, L., Pethick, S., Renda, C., Cox, J. L., Dale, P. S., & Reznick, J. S. (2000). Short-form versions of
the MacArthur Communicative Development Inventories. Applied Psycholinguistics, 21, 95–115.
Fivush, R. (1989). Exploring sex differences in the emotional content of mother–child conversa-
tions about the past. Sex Roles, 20, 675–691.
Fivush, R. (1991). Gender and emotion in mother–child conversations about the past. Journal of
Narrative & Life History, 1, 325–341.
Fivush, R., Brotman, M. A., Buckner, J. P., & Goodman, S. H. (2000). Gender differences in
parent–child emotion narratives. Sex Roles, 42, 233–253.
Flory, C. D. (1935). Sex differences in skeletal development. Child Development, 6, 205–212.
Friederici, A. D., Pannekamp, A., Partsch, C.-J., Ulmen, U., Oehler, K., Schmitzler, R., et al.
(2008). Sex hormone testosterone affects language organization in infant brain. Cognitive
Neuroscience and Neuropsychology, 29, 283–286.
Friedman, S. L., & Jacobs, B. S. (1981). Sex differences in neonates’ behavioral responsiveness to
repeated auditory stimulation. Infant Behavior & Development, 4, 175–183.
Gilligan, C. (1982). In a different voice: Psychological theory and women’s development.
Cambridge, MA: Harvard University Press.
Gleason, J. B. (1975). Fathers and other strangers: Men’s speech to young children. In D. P. Data
(Ed.), Developmental psycholinguistics: Theory and application (pp. 289–297). Washington,
DC: Georgetown University Press.
Goldberg, W. A., Tan, E. T., & Thorsen, K. L. (2009). Trends in academic attention to fathers,
1930–2006. Fathering, 7, 159–179.
Golombok, S., & Fivush, R. (1994). Gender development. New York: Cambridge University Press.
Gray, J. (1992). Men are from Mars, women are from Venus: A practical guide for improving com-
munication and getting what you want in your relationships. New York: Harper Collins.
Greenspan, S. I. (1988). ‘The second other’: The role of the father in early personality formation and
the dyadic-phallic phase of development. In S. H. Cath, A. R. Gurwitt, & J. M. Ross (Eds.), Father
and child: Developmental and clinical perspectives (pp. 123–138). New York: Basil Blackwell.
Halari, R., Hines, M., Kumari, V., Mehrotra, R., Wheeler, M., Ng, V., et al. (2005). Sex differ-
ences and individual differences in cognitive performance and their relationship to endogenous
gonadal hormones and gonadotropins. Behavioral Neuroscience, 119, 104–117.
Halari, R., Sharma, T., Hines, M., Andrew, C., Simmons, A., & Kumari, V. (2006). Comparable
fMRI activity with differential behavioral performance on mental rotation and overt verbal flu-
ency tasks in healthy men and women. Experimental Brain Research, 169, 1–14.
Halpern, D. F. (2002). Sex differences in cognitive abilities (3rd ed.). Hillsdale, NJ: Lawrence Erlbaum.
Hampson, E., van Anders, S. M., & Mullin, L. I. (2006). A female advantage in the recogni-
tion of emotional facial expressions: Test of an evolutionary hypothesis. Evolution and Human
Behavior, 27, 401–416.
Hanlon, H. W., Thatcher, R. W., & Cline, M. J. (1999). Gender differences in the development of
EEG coherence in normal children. Developmental Neuropsychology, 16, 479–506.
Hittelman, J. H., & Dickes, R. (1979). Sex differences in neonatal eye contact time. Merrill-Palmer
Quarterly, 25, 171–184.

Downloaded from fla.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016


24 First Language XX(X)

Holdgrafer, G. E. (1991). Quantity of communicative behavior in children from birth to 30 months.


Perceptual & Motor Skills, 72(3, Pt 1), 803–806.
Howell, D. C. (1992). Statistical methods for psychology. Belmont, CA: Duxbury Press.
Huttenlocher, J., Haight, W., Bryk, A., Seltzer, M., et al. (1991). Early vocabulary growth: Relation
to language input and gender. Developmental Psychology, 27, 236–248.
Jansz, J. (2000). Masculine identity and restrictive emotionality. In A. H. Fischer (Ed.), Gender and
emotion: Social psychological perspectives (pp. 166–186). New York: Cambridge University Press.
Jones, L. C., & Heermann, J. A. (1992). Parental division of infant care: Contextual influences and
infant characteristics. Nursing Research, 41, 228–234.
Kalenkoski, C. M., Ribar, D. C., & Stratton, L. S. (2006). Parental child care in single parent,
cohabiting, and married couple families: Time diary evidence from the United States and the
United Kingdom (Working Paper No. 440). Annandale-on-Hudson, NY: The Levy Economics
Institute of Bard College.
Kilmartin, C. T. (2007). The masculine self (3rd ed.). Cornwall-on-Hudson, NY: Sloan Publishing.
Kraft, R. H., & Nickel, L. D. (1995). Sex-related differences in cognition: Development during
early childhood. Learning and Individual Differences, 7, 249–271.
Kuebli, J., Butler, S., & Fivush, R. (1995). Mother–child talk about past emotions: Relations of
maternal language and child gender over time. Cognition & Emotion, 9, 265–283.
Kuebli, J., & Fivush, R. (1992). Gender differences in parent–child conversations about past
emotions. Sex Roles, 27, 683–698.
Lamb, M. E. (1981). The role of the father: An overview. In M. E. Lamb (Ed.), The role of the
father in child development (2nd ed., pp. 1–63). New York: Wiley.
Lamb, M. E., & Oppenheim, D. (1989). Fatherhood and father–child relationships: Five years
of research. In S. H. Cath, A. Gurwitt, & L. Gunsberg (Eds.), Fathers and their families
(pp. 11–26). Hillsdale, NJ: Analytic Press.
Leaper, C., Anderson, K. J., & Sanders, P. (1998). Moderators of gender effects on parents’ talk to
their children: A meta-analysis. Developmental Psychology, 34, 3–27.
Leaper, C., & Gleason, J. B. (1996). The relationship of play activity and gender to parent and child
sex-typed communication. International Journal of Behavioral Development, 19, 689–703.
Leaper, C., & Smith, T. E. (2004). A meta-analytic review of gender variations in children’s lan-
guage use: Talkativeness, affiliative speech, and assertive speech. Developmental Psychology,
40, 993–1027.
Leeb, R. T., & Rejskind, F. G. (2004). Here’s looking at you, kid! A longitudinal study of perceived
gender differences in mutual gaze behavior in young infants. Sex Roles, 50, 1–5.
Levant, R. F. (1996). The crisis of connection between men and women. Journal of Men’s Studies,
5, 1–12.
Levant, R. F., Good, G. E., Cook, S. W., O’Neil, J. M., Smalley, K. B., Owen, K., et al. (2006). The
normative male Alexithymia Scale: Measurement of a gender-linked syndrome. Psychology of
Men & Masculinity, 7, 212–224.
Levant, R. F., Richmond, K., Majors, R. G., Inclan, J. E., Rossello, J. M., Heesacker, M., et al.
(2003). A multicultural investigation of masculinity ideology and alexithymia. Psychology of
Men & Masculinity, 4, 91–99.
Levy, J., & Heller, W. (1992). Gender differences in human neuropsychological function. In
A. A. Gerall, H. Moltz, & I. L. Ward (Eds.), Sexual differentiation: Handbook of behavioral
neurobiology (Vol. 11, pp. 245–274). New York: Plenum Press.

Downloaded from fla.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016


Lovas 25

Lewis, C., & Gregory, S. (1987). Parents’ talk to their infants: The importance of context. First
Language, 7, 201–216.
Lovas, G. S. (2005). Gender and patterns of emotional availability in mother–toddler and father–
toddler dyads. Infant Mental Health Journal, 26, 327–353.
Maccoby, E. E. (1990). Gender and relationships: A developmental account [American
Psychological Association: Distinguished Scientific Contributions Award Address, 1989, New
Orleans, Louisiana]. American Psychologist, 45, 513–520.
Mannle, S., & Tomasello, M. (1987). Fathers, siblings, and the bridge hypothesis. In K. E. Nelson &
A. van Kleeck (Eds.), Children’s language (Vol. 6, pp. 23–41). Hillsdale, NJ: Lawrence Erlbaum.
Masur, E. F. (1987). Imitative interchanges in a social context: Mother–infant matching behavior
at the beginning of the second year. Merrill-Palmer Quarterly, 33, 453–472.
Masur, E. F., & Gleason, J. B. (1980). Parent–child interaction and the acquisition of lexical
information during play. Developmental Psychology, 16, 404–409.
Maume, D. J. (2008). Gender differences in providing urgent childcare among dual-earner fami-
lies. Social Forces, 87(1), 273–297.
McLaughlin, B., White, D., McDevitt, T., & Raskin, R. (1983). Mothers’ and fathers’ speech to
their young children: Similar or different? Journal of Child Language, 10, 245–252.
Milkie, M. A., Bianchi, S. M., Mattingly, M. J., & Robinson, J. P. (2002). Gendered division
of childrearing: Ideals, realities, and the relationship to parental well-being. Sex Roles,
47, 21–38.
Miller, J. F. (1981). Assessing language production in children: Experimental procedures.
Baltimore, MD: University Park Press.
Notman, M. T., & Nadelson, C. C. (1991). A review of gender differences in brain and behavior.
In M. T. Notman & C. C. Nadelson (Eds.), Women and men: New perspectives on gender
differences (pp. 23–34). Washington, DC: American Psychiatric Press.
O’Brien, M., & Nagle, K. J. (1987). Parents’ speech to toddlers: The effect of play context. Journal
of Child Language, 14, 269–279.
O’Neil, J. M., Good, G. E., & Holmes, S. (1995). Fifteen years of theory and research on men’s
gender role conflict: New paradigms for empirical research. In R. F. Levant & W. S. Pollack
(Eds.), A new psychology of men (pp. 164–206). New York: Basic Books.
Olafsen, K. S., Ronning, J. A., Kaaresen, P. I., Ulvund, S. E., Handegard, B. H., & Dahl, L. B.
(2006). Joint attention in term and preterm infants at 12 months corrected age: The signifi-
cance of gender and intervention based on a randomized controlled trial. Infant Behavior &
Development, 29, 554–563.
Pan, B. A., Rowe, M. L., Spier, E., & Tamis-LeMonda, C. (2004). Measuring productive vocabu-
lary of toddlers in low-income families: Concurrent and predictive validity of three sources of
data. Journal of Child Language, 31, 587–608.
Paquette, D., Bolte, C., Turcotte, G., Dubeau, D., & Bouchard, C. (2000). A new typology of
fathering: Defining and associated variables. Infant and Child Development, 9, 213–230.
Pollack, W. S. (2005). ‘Masked men’: New psychoanalytically oriented treatment models for adult
and young adult men. In G. E. Good & G. R. Brooks (Eds.), The new handbook of psycho-
therapy and counseling with men: A comprehensive guide to settings, problems, and treatment
approaches (rev. ed., pp. 203–216). San Francisco, CA: Jossey-Bass.
Reinisch, J. M., & Sanders, S. A. (1992). Prenatal hormonal contributions to sex differences in
human cognitive and personality development. In A. A. Gerall, H. Moltz, & I. L. Ward (Eds.),

Downloaded from fla.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016


26 First Language XX(X)

Sexual differentiation: Handbook of behavioral neurobiology (Vol. 11, pp. 221–243). New
York: Plenum Press.
Ring, E. D., & Fenson, L. (2000). The correspondence between parent report and child perfor-
mance for receptive and expressive vocabulary beyond infancy. First Language, 20, 141–159.
Roche, A. F. (1968). Sex-associated differences in skeletal maturity. Acta Anatomica, 71, 321–340.
Rondal, J. A. (1980). Fathers’ and mothers’ speech in early language development. Journal of
Child Language, 7, 353–369.
Rowe, M. L., Coker, D., & Pan, B. A. (2004). A comparison of fathers’ and mothers’ talk to
toddlers in low-income families. Social Development, 13, 278–291.
Sayer, L., Bianchi, S., & Robinson, J. P. (2004). Are parents investing less in children? Trends in
mothers’ and fathers’ time with children. American Journal of Sociology, 110, 1–43.
Shepard, D. S. (2002). A negative state of mind: Patterns of depressive symptoms among men with
high gender role conflict. Psychology of Men & Masculinity, 3, 3–8.
Shucard, D. W., Shucard, J. L., & Thomas, D. G. (1987). Sex differences in the patterns of scalp-
recorded electrophysiological activity in infancy: Possible implications for language develop-
ment. In S. U. Philips & S. Steele (Eds.), Language, gender, and sex in comparative perspective
(pp. 278–295). New York: Cambridge University Press.
Silverman, D. K. (1987). Female bonding: Some supportive findings for Melanie Klein’s views.
Psychoanalytic Review, 74, 201–215.
Stolt, S., Haataja, L., Lapinleium, H., & Lehtonen, L. (2008). Early lexical development of Finnish
children: A longitudinal study. First Language, 28, 259–279.
Tannen, D. (1990). You just don’t understand: Women and men in conversation. New York: Ballantine.
Tenenbaum, H. R., & Leaper, C. (1998). Gender effects on Mexican-descent parents’ questions and
scaffolding during toy play: A sequential analysis. First Language, 18, 129–147.
Van Hooren, S. A. H., Valentijn, A. M., Bosma, H., Ponds, R. W. H. M., van Boxtel, M. P. J., &
Jolles, J. (2007). Cognitive functioning in healthy older adults aged 64–81: A cohort study into
the effects of age, sex and education. Aging, Neuropsychology, and Cognition, 14, 40–54.
Volling, B. L., & Belsky, J. (1992). The contribution of mother–child and father–child relationships
to the quality of sibling interaction: A longitudinal study. Child Development, 63, 1209–1222.
Walker, K., & Armstrong, L. (1995). Do mothers and fathers interact differently with their child or
is it the situation which matters? Child: Care, Health & Development, 21, 161–181.
Wong, Y. J., Pituch, K. A., & Rochlen, A. B. (2006). Men’s restrictive emotionality: An investiga-
tion of associations with other emotion-related constructs, anxiety, and underlying dimensions.
Psychology of Men & Masculinity, 7, 113–126.

Downloaded from fla.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016

Вам также может понравиться