Send Orders of Reprints at email@example.com
The Open Communication Journal, 2013, 7, 1-11 1
1874-916X/13 2013 Bentham Open
Speaking the Language of Love: On Whether Chapman’s (1992) Claims
Stand Up to Empirical Testing
Denise M. Polk* and Nichole Egbert
Main Hall, West Chester University, West Chester, PA 19383, USA
Abstract: This paper explores and tests the claims made by Chapman (1992) in his popular press book, The Five Love
Languages: How to Express Heartfelt Commitment to your Mate. One of Chapman’s fundamental claims is that couples
where partners receive their respective preferred love languages experience higher quality relationships than couples who
do not. Couples (N = 83) reported their preferences for and tendencies to demonstrate Chapman’s five love languages.
They also completed measures of relational quality. Descriptive results revealed different potential couple combinations in
terms of feeling and giving preferred love languages, and suggest that few couples meet Chapman’s criteria for high
relational quality. After collapsing couple combinations to reflect matched, mismatched, or p artially matched couples (in
terms of feeling and giving their love language preferences), a significant result surfaced regarding assessments of
relational quality. More specifically, matched and mismatched couples’ reports of relational quality exhibited less
discrepancy th an partially matched couples. Other results from tests of ANOVA and MANOVA provided little empirical
support for Chapman’s notions of love languages.
Keywords: Love languages, relational quality, affection, romantic couples.
“He sends you flowers when what you really want is time
to talk… The problem isn’t your love – it’s your love
language” (Chapman, 1992, back cover). In his best-selling
book, “The Five Love Languages: How to Express Heartfelt
Commitment to Your Mate,” Dr. Gary Chapman promoted a
theory that has gained widespread public approval. For
example, the government of Singapore and the Chaplain’s
Office of NATO invited him to speak, and the book has been
a perennial New York Times Bestseller, selling over seven
million copies (Marriage & Family Life Consultants).
Chapman’s main thesis is that there are five emotional love
languages (LLs) – ways that people “speak” and understand
emotional love. Despite the fact that the number of ways to
express love through LLs is essentially limitless, people
must learn to “speak” the LL of their partner because
relational satisfaction hinges on filling a partner’s
metaphorical emotional “love tank” (Chapman, 1992).
Academic researchers often criticize popular books as
oversimplifying complex ideas, but Chapman’s (1992)
claims parallel some relational scholarship. For example,
Egbert and Polk (2006) found the LLs formed five distinct
factors, and they found significant relationships between
several relational maintenance factors and LLs. Therefore,
one goal of this project is to test the foundation of
Chapman’s claims through empirical investigation.
Specifically, we tested Chapman’s thesis that couples where
partners tend to give love in ways that aligns with their
*Address correspondence to this author at the Main Hall, West Chester
University, West Chester, PA 19383, USA; Tel: 610-436-2658;
Fax: 610-436-3046; Emails: firstname.lastname@example.org
partners’ preferred LLs actually enjoy higher quality
CHAPMAN’S FIVE LOVE LANGUAGES
Empirical support for Chapman’s (1992) book is
mixed when compared with communication scholarship.
Chapman’s (1992) basic claim about the fundamental need
for love and affection is well-documented empirically
(Floyd, 2006; Floyd, Hesse, & Haynes, 2007; Schutz, 1958),
affecting well-being (Downs & Javidi, 1990) and affecting
different types of relationships (e.g., Floyd & Morman,
2003; Schrodt, Ledbetter, & Ohrt, 2007). It also plays a
significant role in relational maintenance (Bell & Healey,
1992) and quality (Floyd & Morman, 1998). In fact, Floyd
(2006) claimed that humans need to be shown they are
loved, and other researchers have documented ways people
accomplish this expression (Villard & Whipple, 1976).
Several theories can be used to predict and/or explain
affectionate behaviors. For example, Floyd (2006) reviewed
theories used to frame studies of affectionate behavior, and
included Thibaut and Kelley’s (1959) interdependence
theory, expectancy violations theory (Burgoon, 1978), and
interaction adaptation theory (Burgoon, Dillman, & Stern,
1993). Floyd’s review explained how affection exchange
theory (Floyd, 2001) addresses the existing theories’
inability to predict or explain affectionate communication.
According to Chapman (1992), no emotional need is
more basic than the need for love and affection, and people
express love according to five LLs: words of affirmation
(encouraging and affectionate messages), quality time (spent
together relating or in shared activities), gifts (thoughtful
tokens), acts of service (help with tasks), and physical touch
2 The Open Communicatio n Journal, 2013, Volume 7 Polk and Egbert
(hand-holding to sexual intercourse). Although Chapman
uses the term “speak,” four of these LLs largely are
nonverbal; however, despite the word choice, the five LLs
include behaviors that fall under the scope of what Floyd and
Morman (1998) named a tripartite model of affectionate
behavior (verbal, nonverbal, and supportive behaviors).
The first part of Chapman’s (1992) thesis is that people
tend to have a distinct preference for a specific LL. Chapman
claimed that, often, people will instantly know their own LL
after hearing it described. Similarly, Hazan and Shaver
(1994) successfully tested adult romantic attachment by
collapsing scale items into forced-choice items; yet, more
recently, researchers advocate using continuous measures
(e.g., Feeney, 2008). Employing Egbert and Polk’s (2006)
validated 20-item Love Language Scale (LLS; four items for
each dimension), our first goal was to compare forced-choice
measurement with the 20-item scale using continuous scores.
Therefore, we asked:
RQ1: Does a relationship exist between a partner’s
forced-choice feel love language and the means associated
with their own feel Love Language Scale subscores?
In the second part of his thesis, Chapman (1992) added
that when partners speak each other’s primary LL, their need
for love will be satisfied, resulting in high relational quality;
however, when they do not, their love tank will drain. He
suggests that receiving a singular preferred LL is more
important for keeping the tank full than receiving a
combination of all five. Yet, empirical data from other
researchers suggests otherwise. For example, social support
literature suggests that support functions differentially and
impacts outcomes based on contextual needs (Cutrona &
Suhr, 1990, 1992). Johnson (2001) suggested that the more
behaviors practiced, the greater the relational satisfaction,
and Leverett (2007) claimed that the relationship between
maintenance and satisfaction may be dependent on quantity
and quality of behaviors.
According to Chapman (1992), couples often have
different LL preferences. This can pose problems because
most people automatically give their own preferred LL
regardless of their partner’s LL preference (Chapman, 1992).
Therefore, a key to high relational quality is to recognize a
partner’s preference and to engage in behaviors that
communicate that particular LL. This claim has been
supported empirically. For example, Floyd (2006) found that
“although affectionate behaviors may carry some inherent
positivity, their valence is also determined by the extent to
which they are congruent with a recipient’s desires” (p. 86).
Thus, people may give the LL they prefer to receive, hoping
it will be reciprocated. However, Floyd (2006) claimed that
people often compensate when they receive affection
incongruent with their desires, or they ignore/fail to perceive
the behavior as affectionate. Chapman echoed these ideas,
claiming that mismatches occur when one partner fails to
recognize and respond appropriately to a partner’s LL. Floyd
(2006) also addressed this issue, arguing that ignoring
affection behaviorally often indicates the recipient is
uncertain as to how to respond. Of course, if both people
desire the same LL, then it is likely that partners tend to
reciprocate that LL increasingly over time, leading to higher
relational quality. Using Floyd’s (2006) language of
affectionate behavior, then, Chapman’s thesis hinges on the
idea that when people’s LLs are different, they should
compensate for those differences by actively choosing
behaviors that reflect the other person’s desired LL.
On the other hand, Chapman’s (1992) ideas diverge from
some scholars about the frequency of affectionate behaviors.
For example, Villard and Whipple (1976) contended that
people ascribe more value to rarely-used currencies, value
frequently used currencies less, and that unused currencies
possess no value. Dainton (2000) found that people expect
partners to perform all types of relational maintenance
behaviors. According to Chapman, however, the frequent
expression of a partner’s LL is the best contribution to
relational quality, and although Chapman agrees that
partners must exchange desired LLs over the long term, he
does not advance that LLs can be equally valued beyond one
(or sometimes two) favorite/s.
Furthermore, people differ regarding their expression and
receipt of affectionate communication (Floyd, 2003, 2006;
Floyd, et al., 2005). Floyd (2006) outlined a range of optimal
tolerance for affection that considers both need and desire, as
well as upper and lower thresholds. These thresholds both
can be problematic in different ways (Floyd, 2006). This is
very different from Chapman’s (1992) claim that only the
failure to receive one’s minimum threshold is problematic.
He does not address the possibility that people potentially
could receive too much affection. To understand people’s
tendency to enact the LL behaviors that they prefer to
receive rather than the LL behaviors their partner prefers, to
receive we ask:
RQ2: Does a relationship exist between a partner’s
forced-choice feel love language preference with their
partner’s tendency to give Love Language Scale subscores?
Chapman (1992) argues that the key to relational quality
is more than recognizing a partner’s LL by learning how to
enact behaviors that demonstrate the LL. He maintains that
people must consciously prioritize a partner’s needs to
enhance relational quality, but does not offer empirical
evidence for his claims, nor does he discuss situations where
both partners receiving their desired LL, only one partner
receiving his or her desired LL, or neither receiving desired
LLs. This information can be obtained by categorizing each
couple based on the LL preferences they report
feeling/preferring and giving. Once all the types have been
identified, they can be collapsed to represent matches (both
report giving each other’s preferred LL), partial matches
(only one reports giving the other’s preferred LL), and
mismatches (neither person reports giving their partner’s
preferred LL). Therefore, the following question provides a
basis for exploring these claims:
RQ3: What are the most common couple types given the
different potential LL configurations?
Moving forward to examine Chapman’s claims about
relational satisfaction, evidence from empirical scholarship
is largely supportive. Burleson and Denton (1992) predicted
that a couple’s similarity in social skill impacts marital
satisfaction. Burleson, Kunkel, and Birch (1994) found that
although similarity in communication did not impact
whether people dated each other, it contributed to their
relational satisfaction and partner attraction. A basic idea
behind interdependence theory (Thibaut & Kelley, 1959) is
Love Languages The Open Communication Journal, 2013, Volume 7 3
that as couples become more deeply involved, they become
more dependent upon their relationship. This dependence is
linked with satisfaction and commitment. More importantly,
the greater their satisfaction and commitment, the more
likely they are to use pro-relationship behaviors (i.e.,
relational maintenance) to preserve and maintain that
satisfaction and commitment (Rusbult, Olsen, Davis, &
This is also true of interaction adaptation theory
(Burgoon et al., 1993) which posits that people compare
their needs, expectations, and desires to the behaviors of
conversational partners and reciprocate behaviors that match
or are more positive than those needs, expectations, and
desires. Floyd and Burgoon (1999) found that people will
match increasing affectionate behavior and compensate for
decreasing affectionate behavior when they desire and expect
affection. They also addressed the outcomes of situations
where people desire one thing but expect another. Chapman
(1992) really does not address the possibility that although
people may desire a particular LL, they might expect their
spouse to give a different one (perhaps simply based on past
interactions). Instead, he argued that people cannot and do
not feel loved if partners do not provide their desired LL,
often because the enacted behaviors may not register as
affectionate behaviors. However, Dainton’s (2000) results
support Chapman’s claim about the relationship between
LLs and satisfaction in that the extent to which expectations
about partner maintenance behaviors were met related
positively to relational satisfaction. Thus, failing to enact
certain behaviors may lead partners to feel unloved.
Chapman’s theory and interdependence theory suggest that
relational quality relates to partners meeting or exceeding
expectations of receiving their desired LL; thus, this study
explored how matches and mismatches in giving/getting LLs
could impact relational quality. Therefore, we asked the
following two research questions:
RQ4: Is self-reported relational quality impacted by the
degree to which one partner’s feel love language preference
matches their partner’s reported tendency to give love
RQ5: Is self-reported relational quality predicted by
MATERIALS AND METHODOLOGY
Participants and Procedures
Couples (N - 86) included students enrolled in a speech
course at a large Midwestern university who also were in a
current romantic relationship of at least two months (n - 95)
and their romantic partners (n - 71) [86 females, 83 males, 3
unreported: ages 18-22 (n - 148), 23-30 (n - 11), 30-40 (n -
3), over 40 (n - 2), and no age reported (n - 8)]. Three
couples were excluded from the study because at least one
partner left a significant number of items blank. Students
received course credit for completing this study (some
couples involved students for both partners), and the names
of non-student partners were put into a gift certificate
drawing for their participation. Students and their romantic
partners completed the questionnaires under the authors’
supervision in a university classroom. Couples arrived
together and received questionnaire packets with
corresponding codes so partners could be matched up.
Participants were predominantly Caucasian (n - 147):
[African American (n - 6), Asian American (n - 1), & “other”
or unreported (n - 18)]. Most participants were first or
second year students (n - 107): [juniors or seniors (n - 35),
graduate students (n - 1), not in college (n - 16) & unreported
(n - 13)]. A majority of participants reported their marital
status as never married (n - 125): [married, divorced, or
widowed (n - 34) & “other” or unreported (n - 13)].
Relationship length ranged from 2-6 months (n - 39), 6
months to 2 years (n - 67), 2-5 years (n - 43), over 5 years (n
- 15), and unreported (n - 8).
Love languages. Participants received a forced-choice LL
measure. The instructions read: “Please select the statement
that best describes you by filling in ONE of the
appropriate/corresponding bubbles. It may be hard to choose
just ONE answer, but try to figure out which of the following
is most important to you... ” Participants had five choices of
“I feel the most love when my partner&”: (1) physically
touches me (i.e., gives a hug, gives a kiss, holds my hand,
touches me), (2) helps me out (i.e., running an errand,
finishing a chore for me, helping me out, helping to keep
things cleaned up), (3) spends quality time with me (i.e.,
really listening, doing something we both like, engages in
quality conversation, spending free time), (4) says
encouraging words (i.e., compliments, expresses
appreciation for me, gives me credit for something I did,
gives me positive comments), or (5) gives me gifts (i.e., a
thoughtful birthday gift, a greeting card, a present for no
special reason, a gift after being away). These items were
collapsed from Egbert and Polk’s (2006) 20-item LLS. This
method parallels Hazan and Shaver (1987, 1994), who
collapsed attachment style scale items into forced-choice
items, one for each style.
Later in the questionnaire packet, each participant also
completed two versions of the LLS – in one version,
participants responded to each item about how they tended to
prefer, or feel, love whereas in the other version, they
responded about how they tended to give love to their
partner. The LLS scale consists of 20-Likert-type items that
represent Chapman’s (1992) five different LLs (four items
for each dimension). Egbert and Polk (2006) reported
sufficient reliability and construct validity, demonstrating
significant relationships between the LLs and relational
The rationale for creating a forced-choice LL and then
also having them complete the LLS was to explore the extent
to which people can self-identify their LL and the extent to
which that preference is reflected in the LLS score when
they could rate all five LLs. Although Hazan and Shaver’s
(1987) method of measuring attachment by selecting a single
statement has been established as consistently reliable (e.g.,
Fuller & Fincham, 1995; Hazan & Shaver, 1994; Meyers &
Landsberger, 2002; Weger & Polcar, 2000; 2002), more
recently researchers advocate continuous measures (Cassidy
& Shaver, 2008).
Reliability analyses from the current study suggested the
LLS is a reliable measure. Cronbach’s alphas ranged from
.80 to .85 for participant responses regarding how they feel
4 The Open Communicatio n Journal, 2013, Volume 7 Polk and Egbert
and give love. These numbers are in line with previous
reliability (see Egbert & Polk, 2006). In addition,
confirmatory factor analyses (CFAs) using AMOS 20.0
helped test each scale’s validity. This also helped to identify
any potentially problematic items that might compound any
reduction to the goodness of fit of the overall model. Results
of the CFAs suggested a good fit for each of the five LL
dimensions: words (
2 - 34.37; df - 19; p < .05; GFI - .95;
RMSEA - .07), time (
2 - 50.61; df - 19; p < .001; GFI - .93;
RMSEA - .09), gifts (
2 - 55.99; df - 19; p < .001; GFI - .93;
RMSEA - .09), touch (
2 - 20.35; df - 19; p - .37; GFI - .97;
RMSEA - .07), and acts (
2 - 36.67; df - 19; p < .01; GFI -
.95; RMSEA - .07). RMSEA fits up to .08 may reasonably
account for error (Browne & Cudeck, 1993) and
MacCullum, Browne, and Sugawara (1996) claimed fits of
.08 to .10 represent mediocre fits. In addition, traditionally
an omnibus cut-off point of 0.90 has been recommended for
the GFI; however, when sample sizes are low a higher value
of 0.95 is preferred (Miles & Shevlin, 1998). To avoid
accepting misspecified models, Hu and Bentler (1999)
recommended not accepting values under 0.90. Furthermore,
our sample was small (i.e., defined as less than 200.
Therefore, where small samples are used, the chi-square may
not discriminate between good fitting models and poor
fitting models (Kenny & McCoach, 2003). Researchers have
sought alternative indices to assess model fit. One such
alternative is Wheaton, Muthen, Alwin, and Summer’s
(1977) relative/normed chi-square (χ2/df). Although no
consensus exists about an acceptable ratio for this statistic
(Bollen, 1989), recommendations range from as high as 5.0
(Mueller, 1996; Wheaton et al., 1977) to below 3.00
(Mueller, 1996). Taken together, although the results of the
CFA do not meet all the criteria of the most stringent
guidelines, they certainly do fall within ranges considered
acceptable to good.
Quality of relationships inventory (QRI). Participants
completed the three-dimensional (depth, support, and
conflict), 25-item, Likert-type QRI (1 = not at all to 5 = very
much; Pierce, 1994). The QRI is a valid and reliable
indicator of relational quality, consistently highly correlated
to observers’ ratings of social behavior (Pierce, 1994). In the
current study, reliability of the subscales (Cronbach alphas)
was acceptable: depth = .76 (it contains the fewest items, and
deleting any items further reduced the alpha level), support =
.81, and conflict = .87. These numbers are similar to
Verhofstadt, Buysse, Rosseel, and Peene, (2006) who tested
the psychometric properties of the QRI, separating scores by
gender. Subscale alphas ranged from .79-.88. CFA results
indicated each dimension fit the data (support: χ2- 17.66, df -
14, p - .13, RMSEA - .05, CFI - .97, IFI - .97; depth: χ2 -
16.17, df - 9, p - .06, RMSEA - .07, CFI - .97, IFI - .97;
conflict χ2 - 107.43, df - 54, p - .00, RMSEA - .08, CFI - .92,
IFI - .92). As with the LLS, these numbers do not reflect an
ideal fit, but they suggest what many researchers consider
acceptable levels of fit.
Among male participants, frequencies for forced-choice
LL are as follows: (a) touch feel/give n - 39, 34; (b) acts
feel/give n - 4, 3; (c) time feel/give n - 31, 40; (d) words n -
6, 5; and (e) gifts n - 5, 5. In addition, of the female
participants, self-reports of each of the types of LL are as
follows for feel/give: (a) touch n - 31, 30; (b) acts n - 3, 3;
(c) time n - 33, 35; (d) words n - 14, 12; and (e) gifts n - 2, 3.
Undistinguished gender couples (where one or both partners
did not identify gender (n - 3) were not excluded from the
To address the first question about the forced-choice
option as compared with LLS scores, we ran five separate
one-way ANOVAs for each partner using the participant’s
forced-choice feel LL and the mean scores of their own feel
LLS responses for the five dimensions. Not only were there
no significant differences (see Table 1), but the forced-
choice LL for each individual did not always correspond
with the highest mean score of the five LLS dimensions.
Thus, stating a preference for “touch” in the forced-choice
question did not significantly correspond with higher scores
in the touch subscale of the LLS, as compared with those
who stated a preference for one of the other categories.
Participants’ forced-choice feel LL matched only six of the
highest means of the LLS dimensions (4 of 5 for men and 2
of 5 for women). For men, the LLS mean score for acts was
(M - 17.25, sd - 2.22), for gifts was (M - 19.33, sd - 1.15), for
time was (M - 17.03, sd - 2.52), and for words was (M -
17.67, sd - 1.75). The highest mean for the forced-choice LL
as compared with the LLS means with forced-choice touch
actually was time (M - 18.38, sd - 2.04). For women, forced-
choice answers only corresponded with two of the means for
the LLS dimensions: gifts (M - 19.00, sd - 1.41) and touch
(M - 18.71, sd - 1.85).
As an additional way to address the first question about
the feel forced-choice option as compared with feel LLS
scores, we conducted a mixed-model MANOVA with
forced-choice feel LL as the between subjects factor, and the
feel LLS responses for each of the five dimensions as the
dependent variables. To account for within-dyad variance,
we treated role in dyad (male or female) as a within-subjects
factor. Because role in dyad was input as a within-subjects
factor, we could not also run it as a predictor, between-
subjects factor. Therefore, the results reported represent all
participants and are not broken down by into male and
female. Forced-choice feel LL was not significantly related
to scores on the five feel LLS dimensions: words F(4, 160)-
.49, p-ns, time F(4, 160) - .36, p-ns, gift F(4, 160) - .48, p-ns,
acts F(4, 160) - .51, p-ns, and touch F(4, 160) - .31, p-ns.
Means and standard deviations are reported in Table 2. Thus,
stating a preference for “touch” in the forced-choice question
did not significantly correspond with higher scores in the
touch subscale of the LLS, as compared with those who
stated a preference for another category.
Similarly, to address the second question about the extent
to which partners’ reports of the LLs they tend to give
compared with their partner’s self- identified feel LL, we
conducted another set of five one-way ANOVAs. Again, no
significant differences occurred for individuals (see Table 3).
Descriptively as a group, males’ forced-choice reports
corresponded with the highest means for the LLS only on the
acts dimension (M - 18.33, sd - 2.89). For females, only
forced-choice words (M - 18.33, sd - 2.08) and gifts (M -
17.33, sd - 2.52) corresponded with the highest means for the
As an additional way to address the second question
about the extent to which partners’ reports of the LLs they
Love Languages The Open Communication Journal, 2013, Volume 7 5
Table 1. One-Way ANOVAs Comparing Partners’ Forced-Choice Feel Love Language Preferences with their Own Love Language
tend to give compared with their partner’s feel LL, we
conducted another mixed-model MANOVA with forced-
choice feel LL as the between-subjects factor, and the partner
give LLS responses for each of the five dimensions as the
dependent variables. To account for within-dyad variance,
we treated role in dyad (male or female) as a within-subjects
Again, because role in dyad was input as a within-
subjects factor, we could not also run it as a predictor,
between-subjects factor. Therefore, the results represent all
participants and are not broken down by sex. Forced-choice
feel LL was not significantly related to partner give scores on
the five LLS dimensions: words F(4, 160) - 1.43, p-ns, time
F(4, 160) - .54, p-ns, gift F(4, 160) - .07, p-ns, acts F(4,
160)-.40, p-ns, and touch F(4, 160) - 1.02, p-ns. Means and
standard deviations are reported in Table 4.
To address the third research question about the potential
couple types, first it was necessary to record all the different
combinations of couples possible with regard to giving and
feeling LLs. For the purpose of this study, we set parameters
6 The Open Communicatio n Journal, 2013, Volume 7 Polk and Egbert
for couple types based on the forced-choice LL the dyadic
partners reported feeling and giving, resulting in 12 different
types of couples (see Table 5 for couple types, frequencies,
and examples). Then we collapsed those 12 types down into
3 couple types based on whether the partners matched on
giving one another’s felt LL, whether they were partly
matched (one received his/her felt LL but the other did not),
or whether they were mismatched (both partners gave a
different LL than their partner’s felt LL). Couple types 1 and
12 (see Table 5) represent matches-- the couples Chapman
(1992) claimed experience the highest relational quality
(although no couple actually surfaced as Type 12). Couple
types 3, 4, 10, and 11 involve partial matches, and types 2, 5,
6, 7, 8, and 9 involve neither partner receiving his/her
preferred LL. The most frequently occurring couple type
(Type 2, n - 39) represented a mismatch. The within-subject
LL consistency between a partner’s feel preference and what
that partner gives also is worth reporting. For male
participants, 68 reported a match, whereas 15 reported a
partial match or mismatch. For female participants, 72
reported a match as compared with 11 who reported a partial
match or mismatch.
Table 2. Means and Standard D eviations from Mixed-Model
MANOVA of Forced-Choice Feel Love Language
Preferences with Participants’ Own Love Language
Investigating the fourth research question involved
testing Chapman’s prediction that couples who give and
receive one another’s preferred LL experience enhanced
relational quality. We were interested in these collapsed
couple types -- whether couples were matched, partially
matched, or mismatched with their felt LL preferences and
tendencies to give LLs as opposed to whether specific
differences in the LL combinations contribute differently to
relational quality. Therefore, we used the collapsed couple
types of: match (both partners gave and received preferred
LLs; n = 22), partial match (one partner received his/her
preferred LL, but the other did not; n = 13), and mismatch
(neither partner received his/her preferred LL; n = 48). See
Table 6. Then, we conducted a one-way ANOVA. Results
yielded a significant difference for couple combination on
relational quality discrepancy (F (2, 80) = 5.92, p < .005, η2
= 0.13). A Tukey Post Hoc analysis revealed that matched
and mismatched couples report greater consistency (less
discrepancy) than partially matched couples in assessments
of quality. No difference surfaced between matched and
mismatched couples, but both were different than partially
matched couples (partial and match = 11.36, partial and
mismatch = 10.25, match and mismatch = 1.11 (p < .05 with
unequal cell means, not weighted).
For our final question, we wanted to explore the reports
of relational quality (summed QRIs that included both
partners) as to whether it could be predicted by collapsed
couple type (matched, partially matched, or mismatched
couples). We conducted a one-way ANOVA, but the result
was not significant (F (2, 80) - 0.70, p - .50). To understand
more, we examined the total individual relational quality
score for each member of each couple. Then, we placed each
partner into a low or high quality category and compared
whether the couple matched on having high scores, low
scores, or mismatched scores (one low score and one high
score). We then ran a crosstabs analysis. The result was not
significant. χ2 (4, N - 83) - 1.84, p - .76 (see Table 7).
The results of this study extend Chapman’s (1992) thesis
in a few key ways. First, this study tested the construct
validity of Egbert and Polk’s (2006) LLS. The forced-choice
items were based the LLS, so it provided a different way for
testing the predictive ability of the LLS. Because Egbert and
Polk only tested how one’s partner tends to feel loved, this
study extends the validity of the LL items because it
examined both partners, providing a better snapshot how
LLs impact relational quality but suggesting some potential
problems with the concept of LLs.
For first two research questions about the forced-choice
option as compared with the 20-item LLS, results indicated
no significant differences. People’s forced-choice preferred
LL did not surface as a single preferred LL on the 20-item
LLS. The same held true with regard to each partner’s
forced-choice preferred LL and the 20-item LLS for what
partner’s reported tending to give – no significant differences
surfaced. The lack of significant findings suggests further
testing should be conducted before making any
generalizations about LLs. Whereas, it might be easy simply
to say that the LLS did not accurately predict a person’s
preferred LL, we suggest other reasons should be considered.
The main reason we contend that the scale itself is valid
relates to the CFA results for the LLS. The data fit the
models well, or at acceptable levels, especially considering
the sample size. For example one reason the LLS scale did
not accurately predict a person’s preferred LL may relate to
the age of participants. Young romantic couples may
struggle, for example, about how much touch, especially
sexual touch, should define the relationship. This might
indicate that people need a period of time after entering
adulthood before they experience that immediate LL
recognition that Chapman (1992) discussed. Perhaps instead,
at that relational stage, young couples feel that all the LL
behaviors are important. This also would support Dainton’s
(2000) claim that people expect partners to perform all types
of relational maintenance behaviors. Further testing could
debunk Chapman’s notion of a single LL preference and
provide more support for the idea that people expect a
variety of behaviors that do not fall into one particular
category, or LL.
After clustering the couples into 12 types based on
preferences to receive and tendencies to give LLs and then
paring down those 12 types into 3 categories (match, partial
match, or mismatch) in order to address the fourth question,
these data revealed some important findings about the nature
of LLs and how they support Chapman’s (1992) claims. Of
Love Languages The Open Communication Journal, 2013, Volume 7 7
Table 3. One-Way ANOVAs Comparing Partners’ Forced-Choice Feel Love Language with Their Partners’ Tendency to Give
Love Language Scale Subscores
Forced Choice LL
the 83 couples, 22 represent a matching type, 13 represent a
partial match, and 48 represent a match. These findings
suggest Chapman was onto something about partners often
not giving one another’s preferred LL. Given the large
number of mismatches (partial or total) -- a full 73.5% of the
couples experienced a partial or total mismatch as compared
with 26.5% matches – this result points to the possibility that
mismatches could be important to understanding relational
outcomes, and especially that mismatches may negatively
affect relational quality.
Some of the specific findings about the 12 couple types
also are worth discussion. None of the 83 couples matched
the style that Chapman (1992) advocated (needing to learn to
express a partner’s LL). Of the 22 matched couples, none of
them actually had to alter their behavior to match their
partner’s LL – it already matched their own. This finding is
8 The Open Communicatio n Journal, 2013, Volume 7 Polk and Egbert
important because couples might not be adapting to a
partner’s LL preference, or it is an infrequent occurrence
(under 1.2% of couples in this study). The couples that
reported a match were couples where both partners felt and
gave the same LL, suggesting that making a conscious
choice to “speak” a partner’s LL may not apply to the
couples in this study.
Table 4. Means and Standard Deviations from Mixed-Model
MANOVA Comparing Partners’ Forced-Choice
Feel Love Language with Their Partners’ Tendency
to Give Love Language Scale Subscores
Because Chapman (1992) claimed that mismatches often
occur, noting the types of mismatch is worthwhile. Of
mismatched couples, the most frequent and most obvious
finding is that when partners feel different LLs, they are
likely to give what they feel. Our sample included 39
mismatched couples who reported this, and an additional 13
couples reported a partial match, meaning that only one
partner is having his/her love tank filled. This suggests
Chapman is correct in attempting to help people understand
the impact of LL differences, and all of these couple types
are candidates for Chapman’s message.
Perhaps the most interesting couples are the five couple
types (5, 6, 7, 8, and 9; see Table 5) who represented the
other types of mismatches. For example, in couple type 5,
both partners actually prefer the same LL, but neither partner
gives this preferred LL. These partners contradict
Chapman’s (1992) idea that people give their own preferred
LL. In addition, in couple 7, each partner feels a different LL
and gives a LL different from their preference, but it still
does not match each partner’s preference. In addition,
couples 6 and 8 are interesting because they represent a
complex type of couple where one partner gives the LL that
s/he feels (which Chapman claims is natural), and the other
partner gives a LL different from his/her own preference;
however, both partners fail to give the other’s preference.
Luckily, no couples reported the type 9 mismatch where
couples each prefer a different LL from one another; yet,
they give a LL different both from their own and from their
In these mismatched cases, additional variables could
explain the LL discrepancies. First, partners see themselves
as complementary in terms of LL, suggesting they do not
expect the other partner desires the same type of behaviors.
Another explanation is, like Stafford and Canary (1991)
found, the type of relationship (dating, dating seriously,
engaged, married) could factor into reports of relational
behaviors that are similar to LLs. Finally, maybe like Bell,
Daly, and Gonzalez (1987) found, perception of quantity of
behaviors is less important than the type of LL performed.
These data revealed that matched and mismatched
couples reported greater consistency in their individual-level
assessments of relational quality than partially matched
couples. No difference between mismatched and matched
couples arose, but both were different from partially matched
couples. Equity theory provides some support (Walster,
Walster, & Berscheid, 1978); if both partners perceive
similar needs (either being met or unmet) they may perceive
equity. Although not ideal, this situation may be more
satisfactory than when one partner feels underbenefited
whereas the other is receiving what s/he desires. For
example, Sprecher (2001) found underbenefiting, but not
overbenefiting, is significantly associated with distress, and
being underbenefited may motivate people to demand equity
(Hatfield & Rason, 1995). Such demands may play out
differently depending on whether one or both partners feel
underbenefited about LLs.
In addition, Dainton (2003) found inequity was linked
positively with the relational maintenance behavior of
openness and suggested that people may use openness as an
equity restoration behavior. Furthermore, researchers have
linked sexual behaviors (initiating, agreeing to, or refusing
sex) with inequity. Perhaps this means that people change
the LL they give after perceiving inequity in terms of what
they receive – not to punish the partner by discontinuing
his/her felt LL – but by attempting to alert their partner of
The lack of significance for the final research question
may be the result of the effects being washed out by the
combined total scores. The crosstabs analysis results seem
puzzling. Namely, the couples with mismatched LLs largely
reported high relational quality. Perhaps again, as long as
both partners feel underbenefited, they may not experience
diminished relational quality.
LIMITATIONS AND FUTURE DIRECTIONS
One limitation to this study is its homogenous college
student sample. Although this population is socially active
and invested in romantic relationships, the relationships tend
to be less developed than in the general population. In
addition, by virtue of age, many of them tend to be less
experienced in relationships. More diversity of participants
also would make the results more generalizable. Having a
sample that includes couples with a wide scope of
relationship length would help us to verify the extent to
which the LLS accurately measures Chapman’s notion of a
single favorite LL.
Variables like relationship length and age affect
relational behaviors over time. For example, maintenance
changes over the course of relationships (Stafford & Canary,
1991; Canary, Stafford, & Semic (2002). In addition, Ciak,
Hutchison, Reed, and Saner (2009) found that time impacted
people’s attributions of flirting behaviors, Willis and Briggs
(1992) found gender differences in the initiation of touch
among dating or married couples, and Guerrero and
Anderson (1994) found that partners increasingly matched
touch behavior as the relationship developed. Therefore, it is
worthwhile to test the extent to which people’s felt LL
Love Languages The Open Communication Journal, 2013, Volume 7 9
Table 5. Different Love Language Couple Comb inations
Couple matches on both giving and receiving LLs
Partner 1 prefers and gives time;
Partner 2 prefers and gives time
Both partners prefer different LL from each other, and both
partners give their own preferred LL
Partner 1 prefers and gives time; Partner 2
prefers and gives touch
Partner 1 prefers a different LL than gives, and Partner 2
prefers and gives own LL (Partner 1’s LL is same as
Partner 2, so even though Partner 2 gives what Partner 1
wants, there’s no extra effort whereas Partner 1’s giving
involves extra effort)
Partner 1 prefers time but gives touch;
Partner 2 prefers and gives touch
Partner 1 gives own preferred LL, and Partner 2 prefers a
different LL but gives Partner 1’s preferred LL
Partner 1 prefers and gives time; Partner 2
prefers touch but gives time
Both partners prefer the same LL, but both partners do not
give this LL
Partner 1 prefers time but gives touch;
Partner 2 prefers time but gives acts
Partner 1 prefers and gives own LL, but Partner 2 prefers a
different LL and gives Partner 1 a LL different from
Partner 1 prefers touch and gives touch;
Partner 2 prefers time but gives acts
Partner 1 gives a different LL than own preferred but not
matching Partner 2 preference, and Partner 2 gives a
different LL from own preferred but not matching Partner
1 preference; however both partners are giving the same
Partner 1 prefers time but gives touch;
Partner 2 prefers gifts but gives touch
Partner 1 gives a LL different from own preference but not
one that Partner 2 prefers, and Partner 2 gives and prefers
own LL which does not match Partner 1’s preferred
Partner 1 prefers touch but gives acts;
Partner 2 prefers and gives words
Both partners prefer different LLs, and they give LL
different from own preference but not matching partner
Partner 1 prefers touch but gives acts;
Partner 2 prefers gifts and gives words
Partner 1 prefers one LL but gives Partner 2’s preferred
LL, and Partner 2 prefers and gives own LL
Partner 1 prefers touch but gives words;
Partner 2 prefers and gives words
Partner 1 prefers and gives the same LL, but Partner 2
prefers same LL as Partner 1 but gives a different LL
Partner 1 prefers and gives touch; Partner 2
prefers touch but gives acts
Partner 1 prefers one LL but gives Partner 2’s preferred
LL; Partner 2 feels a different LL than Partner 1, but gives
Partner 1’s preferred LL – this is Chapman’s ideal
Partner 1 prefers time but gives touch;
Partner 2 prefers touch but gives time
Note: The examples do not represent all the LL combinations possible within that couple type
changes rather than being trait-like as Chapman (1992)
Future studies also could look at more complex
combinations of data. In addition to collecting data on self-
reports of LL preferences and partner reports of giving LLs,
it would be helpful to gather data on what LL behaviors
people perceive their partners give them. Dindia (2000)
suggested that research should explore the relationship
between perceptions of partner behaviors, satisfaction, and
the amount and type of enacted behaviors. Chapman’s
(1992) theory provides a relevant link to this needed research
because of his clear claims about learning to enact a
partner’s felt LL rather than simply to enact those preferred
by oneself. Perhaps specific LL combinations lead to higher
relational quality. For example, perhaps couples in which
both partners prefer and give time report higher relational
quality than couples in which both partners prefer and give
We hope this research stimulates more study of relational
behaviors and relational quality. Of special note is the
finding that behavioral discrepancies offer more explanatory
power than the behaviors themselves. The most pressing
issue is determining the nature of discrepancies between
partners’ desires and what the other gives. Are these
discrepancies relationship-promoting or the cause for unmet
expectations and disappointment? Whereas this study
10 The Open Communication Journal, 2013, Volume 7 Polk and Egbert
Table 6. Frequency of Collap sed Couple Relational Quality by Collapsed Couple Type
Represented by Couple
Matching High Relational
Matching Low Relational
1 & 12
3, 4, 10, & 11
2, 5, 6, 7, 8, & 9
Table 7. Crosstabs Analysis of Collapsed Couple Relational Quality and Collapsed Couple Type
Quality Match or Mismatch
pointed out the importance of this issue and the possible
configurations of such discrepancies in couples, only future
research can provide the explanations.
CONFLICT OF INTEREST
The authors confirm that this article content has no
conflicts of interest.
The authors wish to thank Mary E. Braz for the
consultation she provided on this project.
Bell, R. A., Daly, J. A., & Gonzalez, M. C. (1987). Affinity-maintenance in
marriage and its relationship to women's marital satisfaction.
Journal of Marriage and Family, 49, 445-454.
Bell, R. A., & Healey, J. G. (1992). Idiomatic communication and
interpersonal solidarity in friends’ relational cultures. Human
Communication Research, 18, 307-355.
Bollen, K. A. (1989). Structural equations with latent variables. New York:
John Wiley & Sons.
Browne, M. W. & Cudeck, R. (1993). Alternative ways of assessing model
fit. In: K. A. Bollen, & J. S. Long, (Eds.) Testing structural
equation models (pp. 136–162). Beverly Hills, CA: Sage
Burleson, B. R., & Denton, W. H. (1992). A new look at similarity and
attraction in marriage: Similarities in social-cognitive and
communication skills as predictors of attraction and satisfaction.
Communication Monographs, 59, 268-287.
Burleson, B. R., Kunkel, A. W., & Birch, J. D. (1994). Thoughts about talk
in romantic relationships: Similarity makes for attraction (and
happiness too). Communication Quarterly, 42, 259-273.
Burgoon, J. K. (1978). A communication model of personal space
violations: Explication and an initial test. Human Communication
Research, 4, 129-142.
Burgoon, J. K., Dillman, L., & Stern, L. A. (1993). Adaptation in dyadic
interaction: Defining and operationalize patterns of reciprocity and
compensation. Communication Theory, 4, 293-316.
Canary, D. J., Stafford, L., & Semic, B. A. (2002). A panel study of the
associations between maintenance strategies and relational
characteristics. Journal of Marriage and the Family, 64, 395–406.
Carmines, E. G., & McIver, J. P. (1981). Analyzing models with unobserved
variables : Analysis of covariance structures. In G. W. Bohrnstedt
and E. F. Borgatta, (Eds.), Social measurement: Current issues (pp.
65-115). Beverly Hills, CA : Sage.
Cassidy, J., & Shaver, P. R. (Eds.) (2008). Handbook of attachment:
Theory, research and clinical applications (2nd ed.). New York:
Chapman, G. (1992). The five love languages: How to express heartfelt
commitment to your mate. Chicago, IL: Northfield Publishing.
Ciak, G., Hutchison, E., Reed, K., & Saner, J. (2009). Same flirting,
different meaning: A study of flirting in casual and serious
romantic relationships. Paper presented at the meeting of the
National Communication Association, Chicago.
Cutrona, C. E., & Suhr, J. A. (1990). Social support and becoming a parent.
In S. Fisher & T. Cooper (Eds.), On the move: The psychological
effects of change and transition (pp. 111-125). New York: Wiley.
Cutrona, C. E., & Suhr, J. A. (1992). Controllability of stressful events and
satisfaction with spouse support behaviors. Communication
Research, 19, 154-174.
Dainton, M. (2000). Maintenance behaviors, expectations for maintenance,
and satisfaction: Linking comparison levels to relational
maintenance strategies. Journal of Social and Personal
Relationships, 17, 827-842.
Dainton, M. (2003). Equity and uncertainty in relational maintenance.
Western Journal of Communication, 67, 164-186.
Dindia, K. (2000). Relational maintenance. In C. Hendrick & S. S. Hendrick
(Eds.), Close relationships: A sourcebook (pp. 287-299). Thousand
Oaks, CA: Sage.
Downs, V. C., & Javidi, M. (1990). Linking communication motives to
loneliness in the lives of older adults: An empirical test of
interpersonal needs and gratifications. Journal of Applied
Communication Research, 18, 32-48.
Egbert, N., & Polk, D. (2006). Speaking the language of relational
maintenance: A validity test of Chapman’s (1992) five love
languages. Communication Research Reports, 23, 1-8.
Feeney, J. A. (2008). Adult romantic attachment: Developments in the study
of couple relationships. In J. Cassidy & P. R. Shaver (Eds.),
Handbook of Attachment: Theory, Research, and Clinical
Applications (pp. 456-481). New York: Guilford.
Floyd, K. (2001). Human affection exchange: I. Reproductive probability as
a predictor of men’s affection with their sons. Journal of Men’s
Studies, 10, 39-50.
Floyd, K. (2003). Human affection exchange: V. Attributes of the highly
affectionate. Communication Quarterly, 50, 135-152.
Love Languages The Open Communication Journal, 2013, Volume 7 11
Floyd, K. (2006). Communicating affection: Interpersonal behavior and
social context. New York: Cambridge University Press.
Floyd, K., & Burgoon, J. K. (1999). Reacting to nonverbal expressions of
liking: A test of interaction adaptation theory. Communication
Monographs, 66, 219-239.
Floyd, K., Hesse, C., & Haynes, M. T. (2007). Human affection exchange:
XV. Metabolic and cardiovascular correlates of trait expressed
affection. Communication Quarterly, 55, 79-94.
Floyd, K., Hess, J. A., Miczo, L. A., Halone, K. K., Mikkelson, A. C., &
Tusing, K. J. (2005). Human affection exchange: VIII. Further
evidence of the benefits of expressed affection. Communication
Quarterly, 53, 285-303.
Floyd, K., & Morman, M. T. (1998). The measurement of affectionate
communication. Communication Quarterly, 46, 144-162.
Floyd, K., & Morman, M. T. (2003). Human affection exchange: II.
Affectionate communication in father-son relationships. Journal of
Social Psychology, 143, 599-612.
Fuller, T. L., & Fincham, F. D. (1995). Attachment style in married couples:
Relation to current marital functioning, stability over time, and
method of assessment. Personal Relationships, 2, 17-34.
Hatfield, E., & Rason, R. L. (1995). A world of passion: Cross-cultural
perspectives on love and sex. New York: Allyn & Bacon.
Hazan, C., & Shaver, P. (1987). Romantic love conceptualized as an
attachment process. Journal of Personality and Social Psychology,
Hazan, C., & Shaver, P. (1994). Attachment as an organizational framework
for research on close relationships. Psychological Inquiry, 5, 1-22.
Hu, L. T. & Bentler, P. M. (1999). Cutoff criteria for fit indexes in
covariance structure analysis: Conventional criteria versus new
alternatives. Structural Equation Modeling, 6(1), 1-55.
Guerrero, L. K., & Anderson, P. A. (1994). Patterns of matching and
initiation: Touch behavior and touch avoidance across romantic
relationship stages. Journal of Nonverbal Behavior, 18, 137-153.
Johnson, A. J. (2001) Examining the maintenance of friendships: Are there
differences between geographically close and long-distance
friendships? Communication Quarterly, 49, 424-435.
Kenny, D. A. & McCoach, D. B. (2003). Effect of the number of variables
on measures of fit in structural equation modeling. Structural
Equation Modeling, 10(3), 333-51.
Leverett, J. E. (2007). Relational maintenance behaviors between college
freshmen and their parents via instant messaging. Master’s thesis.
Retrieved from Proquest on April 18, 2012, from
http://search.proquest.com/docview/304713105 Publication number
MacCallum, R. C., Browne, M. W., & Sugawara, H. M. (1996). Power
analysis and determination of sample size for covariance structure
modeling. Psychological Methods, 1(2), 130-49.
Marriage & Family Life Consultants (n.d.) Biography. Retrieved on August
19, 2011, from http://www.garychapman.org/bio.htm.
Meyers, S. A., & Landsberger, S. A. (2002). Direct and indirect pathways
between adult attachment style and marital satisfaction. Personal
Relationships, 9, 159-171.
Miles, J. & Shevlin, M. (1998). Effects of sample size, model specification
and factor loadings on the GFI in confirmatory factor analysis.
Personality and Individual Differences, 25, 85-90.
Mueller, R. O. (1996). Basic principles of structural equation modeling.
New York: Springer-Verlag.
Pierce, G. R. (1994). The quality of relationships inventory: Assessing the
interpersonal context of social support. In B. Burleson, T. Albrecht,
& I. Sarason (Eds.) Communication of social support: Messages,
interactions, relationships, and community (pp. 247-266).
Thousand Oaks, CA: Sage.
Rusbult, C. E., Olsen, N., Davis, J. L., & Hannon, P. (2001). Commitment
and relationship maintenance mechanisms. In J. H. Harvey & A.
Wenzel (Eds.), Close romantic relationships: Maintenance and
enhancement (pp. 87-113). Mahwah, NJ: Erlbaum.
Schutz, W. (1958). FIRO: A three-dimensional theory of interpersonal
behavior. New York: Rinehart.
Schrodt, P., Ledbetter, A. M., & Ohrt, J. K . (2007). Parental confirmation
and affection as mediators of family communication patterns and
children’s mental well-being. Journal of Family Communication, 7,
Sprecher, S. (200l). Equity and social exchange in dating couples:
Association with satisfaction, commitment, and stability. Journal of
Marriage and Family, 63, 599-613.
Stafford, L., & Canary, D. J. (1991). Maintenance strategies and romantic
relationship type, gender, and relational characteristics. Journal of
Social and Personal Relationships, 8, 217-242.
Thibaut, J. W., & Kelley, H. H. (1959). The Social Psychology of Groups.
New York: John Wiley.
Verhofstadt, L. L., Buysse, A., Rosseel, Y. & Peene, O. J. (2006).
Confirming the three-factor structure of the quality of relationships
inventory within couples. Psychological Assessment, 18, 15-21.
Villard, K., & Whipple, L. (1976). Beginnings in relational communication
(pp. 141-165). New York: John Wiley.
Walster, E., Walster, G. W., & Bershcheid, E. (1987). Equity: Theory and
research. New York: Allyn and Bacon, Inc.
Weger, H., & Polcar, L. E. (2000). Attachment style and the cognitive
representation of communication situations. Communication
Studies, 51, 149-161.
Wheaton, B., Muthen, B., Al win, D., F., & Summers, G. (1977). Assessing
reliability and stability in panel models, Sociological Methodology,
8 (1), 84-136.
Willis, F, N., & Briggs, L .F. (1992). Relationship and touch in pubic
settings. Journal of Nonverbal Behavior, 16, 55-63.
Received: September 20, 2012 Revised: March 19, 2013 Accepted: March 22, 2013
© Polk and Egbert; Licensee Bentham Open.
This is an open access article licensed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/-
licenses/by-nc/3.0/) which permits unrestricted, non-commercial use, distribution and reproduction in any medium, provided the work is properly cited.