Content uploaded by Magdalena Formanowicz
Author content
All content in this area was uploaded by Magdalena Formanowicz on Jan 23, 2018
Content may be subject to copyright.
RESEARCH ARTICLE
Verbs as linguistic markers of agency: The social side of grammar
Magdalena Formanowicz*, Janin Roessel†,CaterinaSuitner‡&AnneMaass‡
* University of Humanities and Social Sciences, Warsaw, Poland; University of Bern, Switzerland
†University of Mannheim, Germany
‡University of Padova, Italy
Correspondence
Magdalena Formanowicz, Department of
Psychology, University of Bern, Fabrikstrasse
8, CH-3012 Bern, Switzerland.
E-mail: magda.formanowicz@gmail.com
Received: 27 October 2015
Accepted: 11 July 2016
http://dx.doi.org/10.1002/ejsp.2231
Keywords: language, meta-semantic
effects, verbs, agency, social judgment
Abstract
Basic grammatical categories may carry social meanings irrespective of
their semantic content. In a set of four studies, we demonstrate that
verbs—a basic linguistic category present and distinguishable in most
languages—are related to the perception of agency, a fundamental dimen-
sion of social perception. In an archival analysis of actual language use in
Polish and German, we found that targets stereotypically associated with
high agency (men and young people) are presented in the immediate
neighborhood of a verb more often than non-agentic social targets
(women and older people). Moreover, in three experiments using a
pseudo-word paradigm, verbs (but not adjectives and nouns) were consis-
tently associated with agency (but not with communion). These results
provide consistent evidence that verbs, as grammatical vehicles of action,
are linguistic markers of agency. In demonstrating meta-semantic effects
of language, these studies corroborate the view of language as a social tool
and an integral part of social perception.
“Yes, we can!”is the most famous political slogan of
our times, propelling many people into action in
favor of then-prospective president Barack Obama.
Interestingly, it is the slogan employing a verb (i.e.,
can) that became the benchmark of Obama’scam-
paign in 2008, rather than terms like “Change”or
“Forward”that were used in parallel. Similarly, the
brand Apple is primarily associated with verb-based
slogans such as “Think Different,”“Get a Mac,”or
“Switch,”rather than slogans without a verb focus,
such as “The Power to Be Your Best.”Why? In this
article, we argue that verbs, as compared with adjec-
tives and nouns, constitute a linguistic category that
conveys information above and beyond the specific
semantic content, and that these meta-semantic effects
influence people’s cognitive processes (see Fiedler,
2008, for a similar notion regarding non-referential
and para-semantic effects of language). Specifically,
we assert that verbs imply dynamic properties that
other grammatical categories (nouns and adjectives)
lack and that make them the preferred syntactic
device to convey activity. By extension, we propose
that verbs also convey agency, a basic dimension in
human perception that is related to goal achieve-
ment and is stereotypically associated with specific
target groups such as men or young people (for an
overview of this notion, see Abele & Wojciszke,
2014). Even if the link between grammatical catego-
ries and agency has (to our knowledge) not
previously been directly examined, the existing
literature yields preliminary support for the notion
of a verb–agency link.
Verbs: The Dynamic Grammatical Category
Until now, the link between grammatical categories
and social meaning was investigated primarily as
regards the concreteness–abstractness dimension
within the Linguistic Category Model (LCM, Semin
& Fiedler, 1988) and its subsequent developments
(Carnaghi et al., 2008). According to this theoretical
perspective, linguistic categories differ in terms of
the abstractness of the information they convey.
Regarding verbs, the LCM distinguishes between
descriptive action verbs, which are the most concrete
and refer to behavior in a specific situation (e.g., “he
kicked the dog”); interpretative action verbs, which
describe a larger class of behaviors (e.g., “he hurt
the dog”); and the third and smallest verb class,
namely state verbs (SV), which describe psychological
states (e.g., “he hates the dog”).
1
Adjectives (e.g., “he
is an aggressive person”)andnouns(e.g.,“he is an
European Journal of Social Psychology 00 (2017) 00–00 Copyright © 2017 John Wiley & Sons, Ltd.
EJSP
aggressor”) are more abstract than verbs and are
mainly used to express general, dispositional judg-
ments (Carnaghi et al., 2008; Gelman & Heyman,
1999; Walton & Banaji, 2004). Whereas nouns and
adjectives represent relatively static, enduring quali-
ties, verbs generally reveal dynamism. Moreover, the
large majority of verbs (descriptive and interpretative
action verbs) convey a sense of agency that is typi-
cally attributed to the subject of the sentence (Brown
& Fish, 1983; Semin & Marsman, 1994). Thus,
already within the LCM framework, verbs are linked
to activity and are seen as mostly denoting actions
and describing behaviors rather than dispositional
traits, for which other grammatical categories become
more distinctive.
On a more basic level, the idea that different word
classes are linked to different meanings also receives
support from neuro-psychological research showing
that prototypical words of different word categories
(such as verbs vs. nouns) recruit partially distinct neural
networks (Caramazza & Hillis, 1991; for a comprehen-
sive overview of brain correlates of grammatical catego-
ries, see Vigliocco, Vinson, Druks, Barber, & Cappa,
2011). However, Vigliocco et al. (2011) noted that clear
neural differences emerged particularly when studies
used prototypical verbs (referring to activities) and pro-
totypical nouns (referring to objects); that is, when a
confound between grammatical classes (verbs vs.
nouns) and their prototypical meanings (action vs. ob-
ject) was present. Yet the observation that verbs proto-
typically coincide with activity is central to our
argument here. Intriguingly, verbs also seem to prompt
muscle activity more than other word classes. For in-
stance, participants’zygomatic major muscle was acti-
vated more strongly when they were exposed to the
verb “to smile”than to the adjective “funny”(Foroni &
Semin, 2009; for a similar pattern of embodied process-
ing of action verbs, see Willems, Hagoort, & Casasanto,
2010).
Taken together, verbs, a basic grammatical category
present in almost all languages (Kroeger, 2005), are lin-
guistic devices used to express actions and agency. The
aim of the present set of studies is to verify whether
the verb–agency relationship extends to the social realm
by testing whether (i) verbs are the preferred word class
when referring to agentic groups, and (ii) whether per-
ceivers correctly infer the agency tied to verbs.
Agency: The Dynamic Dimension in Social
Perception
The “big two,”agency/competence and communion/
warmth, have been identified as the fundamental
dimensions that guide social judgment of the self,
other individuals, and social groups (Abele, Cuddy,
Judd, & Yzerbyt, 2008; Fiske, Cuddy, Glick, & Xu,
2002; Ybarra et al., 2008). By definition, communion
reflects “social acceptance and connection,”whereas
agency allows for the “pursuit of goals, given
available opportunities”(Abele, Cuddy et al., 2008,
p. 1063). Thus, agency (just like verbs), with its focus
on enacting and striving, can be considered the dy-
namic component of the big two.
The dynamic nature of agency is not only reflected in
its content but also in how the construct is structurally
represented in the semantic net. Recent research sug-
gests that communion forms a semantically dense con-
cept (Bruckmüller & Abele, 2013; Fiske et al., 2002),
with closely clustered items (e.g., warm, sociable, and
gentle), whereas agency is perceived as more varied, es-
pecially in its negative components (Bruckmüller &
Abele, 2013; Kenworthy & Tausch, 2008). Similarly,
other research has found that judgments of people’s
agency are easily changed depending on contextual
information (Abele, Rupprecht, & Wojciszke, 2008),
whereas situational changes in communion are rare
(Uchronski, Abele, & Bruckmüller, 2013). On a larger
scale, agentic components of stereotypes change more
easily than communal components (Twenge, 1997).
The above-mentioned research suggests that agency
not only semantically refers to activity, but also repre-
sents a more dynamic construct and is more influenced
by contextual cues, whereas communion is a more
stable and static construct. We further extend the
definition of agency here, showing that not only the
content, or the semantic structure, but also the syntactic
choices contribute to its expression.
Although all of the aforementioned lines of work sug-
gest a link between the grammatical category of verbs
and the social dimension of agency, this link has not
yet been examined directly. In fact, the semantic conno-
tations of agency (vs. communion) have mostly been
examined in reference to adjectives (Abele & Wojciszke,
2007, 2014; Bruckmüller & Abele, 2013). We assert
here that agency is better reflected in verbs than in
adjectives and nouns and that agency concerns doing
actions more than being active. Our core argument is that
agency is reflected in verbs rather than in adjectives and
nouns.
The Present Research
In this article, we address a novel aspect of linguistic cat-
egories by linking verbs to the basic dimension of
agency. The general idea is that verbs are distinctly
1
In a more refined version of the LCM (e.g., Semin & Fiedler, 1991;
Wigboldus & Douglas, 2007), another category is mentioned, namely,
state action verbs (SAV). This is an ambiguous category in the LCM
framework. For instance, in the Linguistic Category Model coding man-
ual (Coenen, Hedebouw, & Semin, 2006—page 7), it is written that
“Because SAVs are very similar to IAVs they are often treated as IAVs.
Moreover, these types of verbs don’t differ significantly in abstraction
level”(Semin & Fiedler, 1991). With the example “Person X thinks
about this topic,”it is also evident that state action verbs convey a sense
of agency and dynamism (a specific topic for a temporary time). This
distinction highlights that action verbs dominate the grammatical cate-
gory of verbs and substantiates our claim that a prototypical verb refers
to an action. We will return to a discussion of state verbs, the smallest
and arguably atypical verb category, in the general discussion.
M. Formanowicz et al.Linguistic markers of agency
European Journal of Social Psychology 00 (2017) 00–00 Copyright © 2017 John Wiley & Sons, Ltd.
associated with agency and that this link is visible both
in language production and in inferences drawn from
language.
Not only is the specific verb–agency link a novel con-
cept, but its meta-semantic nature is as well. We pro-
pose that grammatical categories per se are able to
convey social meaning. This broad claim calls for
tailored research approaches. Until now, only real
words have been used in studies examining the role of
linguistic categories in social psychology. However, in
such cases, the meanings conveyed by the words’
semantics are inevitably involved in the processing of
the grammatical categories. This is to some degree true
even when words share the same stem, but differ in
grammatical category (e.g., to act, active, actor/activist;
to play, playful, player). Even in this case, meaning
may change from one grammatical category to another,
creating a natural confound between the meaning con-
veyed by the grammatical form and by the semantic of
the word. Moreover, there are large frequency differ-
ences with which these words are used (as reflected in
linguistic corpora, an issue we will return to in the gen-
eral discussion). To avoid potential confounds because
of semantics and to approximate the idea of meta-
semantic effects, we employed two complementary
methods for this investigation: (i) a corpora analysis that
investigates word classes beyond their meaning in real
language use and (ii) a pseudo-words paradigm that
excludes the involvement of any semantic content.
Accordingly, we tested the verb–agency link in two
sets of studies. First, we wanted to demonstrate the rela-
tionship between verbs and agency in actual language
use. We therefore decided to turn to existing corpora
and test for co-occurrences of verbs with agentic and
non-agentic social targets (Study 1). The idea of a
verb–agency link for social groups beyond the verbs’
meaning was tested in a series of archival corpora analy-
ses conducted in Polish and German. Because prior re-
search has demonstrated that men are perceived as
more agentic than women (e.g., Diekman & Eagly,
2000; Eagly & Karau, 2002) and young people as more
agentic than old people (e.g., Cuddy, Fiske, & Glick,
2008), we focused on these two pairs of targets. We hy-
pothesized that social targets associated with agency
(i.e., men and the young) would be more likely linked
with verbs than non-agentic targets (i.e., women and
the elderly).
Second, we focused on the interpretation of language
employing a pseudo-word paradigm. We predicted that
verbs (but not adjectives and nouns) are interpreted as
signaling agency; that is, people ascribe more agency
to verbs than adjectives and nouns even when the se-
mantic content is held constant. This hypothesis was
tested in three experiments conducted in Polish using
a pseudo-word paradigm. Pseudo-words with the same
word stem and a suffix unequivocally assigning the
word to the grammatical category of verbs (e.g., to
lann), adjectives (e.g., lannitive), or nouns (e.g.,
lanniness) allowed us to investigate whether grammat-
ical categories intrinsically convey social meaning.
In all experiments (Studies 2 to 4), participants evalu-
ated the degree to which such pseudo-verbs, -adjectives,
and -nouns transmit a sense of agency. In addition to
the focal agency dimension, we also assessed inferences
about communion, the second big two dimension, to
demonstrate discriminant validity. Based on the pro-
posed meta-semantic verb–activity link, we hypothe-
sized that, in contrast to adjectives and nouns, verbs
would be associated with agency, but not communion.
In Studies 3 and 4, we further assessed potential corre-
lates of pseudo-words already investigated in prior re-
search on the big two and/or on language abstraction,
namely valence (Suitner & Maass, 2008) and abstract-
ness (Semin & Fiedler, 1988). Given that we used
pseudo-words free of semantic content, we did not
make predictions about inferences regarding valence
or abstraction. Our main aim was to show that the
verb–agency link will explain unique variance even
when controlling for perceived abstractness and va-
lence. Thus, while the first (corpora) study focuses on
language production, the latter (experimental) studies
investigate the interpretation of words belonging to dif-
ferent word classes and shall establish the distinctness of
the verb–agency link.
Study 1
Are verbs actually used to express agency in spontane-
ous language production? To answer this question, we
investigated the verb–agency link in actual language
use with reference to stereotyped groups. We hypothe-
sized that targets stereotypically characterized by high
agency (men and young people) are more likely associ-
ated with verbs than groups characterized by low
agency (women and old people).
Method
The agency–target association should be strongest when
the target (and not the object) is mentioned as the agent
in the sentence (Fausey & Boroditsky, 2010; Fausey,
Long, Inamori, & Boroditsky, 2010). In exemplary stud-
ies (Fausey & Boroditsky, 2010; Fausey et al., 2010),
when the logical agents occurred in the role of the sub-
ject and were thus paired with the verb (i.e., “he broke
the vase”), they were ascribed more responsibility and
higher financial fines than when the same event was
presented without the crucial agent-verb pairing (“the
vase broke”). Following this logic, we investigated in-
stances in which the target of interest (stereotypically
agentic vs. not) was more often linked with a verb.
Moreover, the target–verb order should enhance the
agency–target association in subject–verb–object
languages such as Polish, and also possibly in languages
without a dominant word order such as German
(Bettinsoli, Maass, Kashima, & Suitner, 2015). There-
fore, we compared instances of verbs directly following
agentic versus non-agentic targets. These sentence
structures should be most common and most effective
in (differentially) conveying agency, which allowed us
M. Formanowicz et al. Linguistic markers of agency
European Journal of Social Psychology 00 (2017) 00–00 Copyright © 2017 John Wiley & Sons, Ltd.
to test our hypothesis while keeping the grammatical
structure constant. In the corpora analyses, we assessed:
(i) the overall occurrence of the target words (to assess
base rates) and (ii) the frequency of the target words
representing high versus low agency immediately
followed by a verb (e.g., a search command for “men
[verb]”instances). To set restrictive criteria for the verbs,
we excluded word forms of the most common auxiliary
and linking words “to have”and “to be”(cf. LCM coding
manual, Coenen et al., 2006) from the analyses (though
the results remain robust when including them).
2
The
study composition is summarized in Table 1.
We recorded the number of occurrences of target–
verb collocations in reference to the total number of tar-
get word occurrences for all search targets. For Polish,
we used the Polish National Corpus (Pęzik, 2012;
http://www.nkjp.uni.lodz.pl). This corpus comprises a
representative sample of language use, consisting of in-
put from books (29%), the press (50%), other written
data (4%), the internet (7%), and spoken language
(10%). For German, we used the archive Tagged-C of
the German Reference Corpus DeReKo (Kupietz,
Belica, Keibel, & Witt, 2010), which comprises over
6.47 million texts of 26 German corpora with approxi-
mately 96% of the texts stemming from newspaper
and press texts and another 3% stemming from the in-
ternet (for a list of all corpora, cf. http://www.ids-mann-
heim.de/cosmas2/projekt/referenz/korpora.html).
3
Results
The results shown in the upper section of Table 2 indi-
cate that in Polish, the co-occurrence of the target group
“men”followed by verbs is 1.38 times more likely than
the co-occurrence of “women”and verbs (odds ra-
tio = 1.45, 95% CI = [1.41; 1.49]; ϕ= 0.06). In German,
this pattern is replicated, with “men”in collocation with
verbs being 1.14 times more likely than “women”in
collocation with verbs (odds ratio= 1.17, 95% CI =
[1.16; 1.18]; ϕ=0.03).
For age, in Polish, the co-occurrence of the target
group “young people”and verbs is 2.89 times more
likely than that of “old people”and verbs (odds ra-
tio = 3.27, 95% CI = [2.73; 3.92]; ϕ= 0.09). In German,
this pattern is replicated, with “young people”in
collocation with verbs being 1.19 timesmore likely than
“old people”in collocation with verbs (odds ratio= 1.22,
95% CI = [1.19; 1.25]; ϕ=0.03).
Discussion
The corpora analyses attest to a verb–agency link in real
language use: stereotypically agentic targets (i.e., men
and young people) were more likely followed by verbs
than non-agentic targets (i.e., women and old people).
This study provides consistent evidence for a natural
link between target agency and verbs in language pro-
duction across a variety of domains and irrespective of
the verbs’meanings. Whether the producers of these in-
numerable instances of language chose their expres-
sions strategically to depict agentic targets as agents
and as being active, or whether they did so unintention-
ally, remains elusive at this point (see also Rubini,
Menegatti, & Moscatelli, 2014, for a detailed discussion
on a motivational approach to language use). For com-
municative functions, however, audiences must be able
to detect the agency hidden in verbs (vs. adjectives and
nouns). Are recipients able to interpret verbs in line
with the (presumed) communicative intentions of the
communicator? To investigate this question, we
conducted three experimental studies.
Experimental Pseudo-Words Paradigm
Studies 2 to 4 were designed to test whether verbs are
ascribed more agency than other linguistic categories.
In order to examine how grammatical categories per
se are related to social meaning, we aimed to “switch
off”the semantic content. To do so, we relied on
pseudo-words tailored to the Polish language. We relied
on Polish for these experiments because in this language
it is possible to indicate the grammatical category based
on the word’ssuffix. Manipulating the grammatical
category would have been more complex in English
and German, for instance, because of the necessity of
adding the indefinite marker “to”in English for a verb
(such as in “to marn”) or dealing with capital letters
and more ambiguous endings in German.
We hypothesized that pseudo-verbs would convey
more agency than pseudo-adjectives and pseudo-
nouns. This effect of grammatical categories should be
specific for the dimension of agency, and it should not
hold for communion judgments.
General Method
4
In all three experiments, participants evaluated a list of
pseudo-words that comprised the same number of
pseudo-verbs, -adjectives, and -nouns (five in Study 2
and three in Studies 3 and 4, respectively). Pseudo-word
sets were selected based on careful pretesting that
2
The results remain robust, however, when including these auxiliary
words in the analyses (Gender Polish: odds ratio = 1.34, 95% CI =
[1.30; 1.37]; Gender German: odds ratio = 1.34, 95% CI = [1.32; 1.35];
Age Polish: odds ratio = 2.24, 95% CI = [1.97; 2.56]; Age German:odds
ratio = 1.25, 95% CI = [1.22; 1.28]).
3
For German, we ran a supplemental analysis with a smaller archive
(Tagged-M), which allowed for more detailed search criteria to maxi-
mize the likelihood that the verb referred to the intended target words.
The verb characteristics were specified as follows: active voice, third or
first person singular or plural(matching the target). Even with these re-
stricted criteria, thetargets characterized by high agency were more fre-
quently followed by verbs than the targets characterized by low agency
(Gender: odds ratio = 1.21, 95% CI = [1.12; 1.31]; Age: odds ratio 1.41,
95% CI = [1.03; 1.92]).
4
The experiments adhere to the APA ethical guidelines and were ap-
proved by an institutional ethics board at the University of Humanities
and Social Sciences (number of the approval 24/IV/11-12).
M. Formanowicz et al.Linguistic markers of agency
European Journal of Social Psychology 00 (2017) 00–00 Copyright © 2017 John Wiley & Sons, Ltd.
assured pronounceability and meaninglessness across
the three linguistic categories. To create the pseudo-
word stems, a formula in Excel randomly generated
phonemes to construe two-syllable word stems
consisting of two C–V–C syllables, whereby C means
consonant and V means vowel. This syllable configura-
tion is among the most frequent in Polish (Śledziński,
2010). The 360 created word stems were screened to
fulfill the following criteria: (i) possible pronunciation
and orthography; (ii) meaninglessness; and (iii) com-
prising biphonemes and triphonemes in the Polish lan-
guage (Śledziński, 2010). In the next step, suffixes
were added indicating either the infinitive of the verb
(-ić,-ować, and -yczyć), the nominative of the adjective
(-ne, -kie, etc.), or the suffix for nouns conveying a
sense of abstractness (-two being typical for words like
manhood—Studies 2 and 3) or suffixes of common
nouns (Study 4). These variants were again screened
to assure ease of pronunciation, orthographic compati-
bility, and absence of meaning.
The final list of 36 pseudo-word sets was pretested
among a group of 26 native Polish speakers. Participants
were randomly assigned to three groups that differed in
the order of presented stimuli (starting with either verb,
adjective, or noun with word stems in the same fixed
random order). Participants were presented with each
of the 36 word stems only once, in the form of either a
verb, adjective, or noun (i.e., each participant viewed
12 verbs, adjectives, and nouns in total). Participants in-
dicated whether they perceived the presented words as
nonsensical (on a scale from 1 = does not at all remind of
an existing word to 6 = reminds very much of an existing
word). Based on these results, 15 pseudo-word sets were
selected so that (i) they were rated similarly in their
nonsensicality across the three linguistic categories
(ps>.05) and (ii) they were significantly different from
the midpoint of the scale (3.5, ps<.05) in the direction
of nonsensicality. The stimuli used in Studies 2 and 3
are presented in the upper part, and those for Study 4,
with partially changed suffixes, are depicted in the
lower part of Table 3.
All studies had a within-subject design with partici-
pants evaluating all types of grammatical categories. To
control for the within-participant variance in the
judgments of grammatical categories, we analyzed the
data using a multilevel framework, with words nested
within participants. Analyses were conducted with
Mplus 7 (Muthén & Muthén, 1998–2012), and the
Maximum Likelihood Robust estimator was used in all
analyses.
Study 2
Method
Participants.Sixty students of mathematics from
Warsaw University participated (31 women, 29 men,
M
Age
=22.23years,SD
Age
=6.24years).
Procedure. For the paper–pencil task, we used the
15 pseudo-word sets obtained from the pretest.
Pseudo-words were presented in a fixed random order,
Tab l e 1. Summary of target stimuli used in Study 1—corpora search with English translations
Target group Language Agentic search targets Non agentic search targets
Gender Men, Man Women, Woman
Polish “Mężczyźni,”“Mężczyzna”“Kobiety,”“Kobieta”
German “Männer,”“Mann”“Frauen,”“Frau”
Age Young People Old people
Polish “Młodzi ludzie,”“Młodzież”“Starzy ludzie,”“Emeryci”
German “Junge Menschen,”“Junge Leute,”
“Jugendliche”
“Alte Menschen,”“Alte Leute,”
“Senioren”
Tab l e 2. Summary of corpora search results (Study 1)
Total With verbs Proportion Odds
Gender
Polish Men 67 460 10 684 0.158 0.188
Women 94 594 10 882 0.115 0.130
χ
2
(1) = 641.01, p<.001
Men 518 321 85 305 0.165 0.197
German Women 675 018 97 326 0.144 0.168
χ
2
(1) = 941.01, p<.001
Age
Polish Young people 20 592 3495 0.170 0.204
Old people 2228 131 0.059 0.062
χ
2
(1) = 185.11, p<.001
German Young peopl e 176 400 26 770 0.152 0.179
Old people 86 519 11 081 0.128 0.147
χ
2
(1) = 264.15, p<.001
Note. Total = frequencyof target words (A);with verbs = frequencyof target words immediately followed by a verb (B); proportion = B / A; odds= B / (A B).
M. Formanowicz et al. Linguistic markers of agency
European Journal of Social Psychology 00 (2017) 00–00 Copyright © 2017 John Wiley & Sons, Ltd.
and participants evaluated lists consisting of 15 stimuli
(five verbs, five adjectives, and five nouns). Three lists
were created, each starting with the same word stem
but a different suffix (i.e., a different grammatical cate-
gory). The instructions read as follows: “This 5 minute
study investigates how meaning is construed in lan-
guage based on artificial grammar. Please evaluate the
following15artificial words. As the words are nonexis-
tent, please rely on your intuition. Please try not to
think for too long, rather rely on your first impression.”
Before participants rated the words, they were asked to
classify them according to their grammatical category
(adjectives, nouns, verbs, other). Overall, 91.56% of
the classifications were correct (for verbs: 99.33%, for
adjectives: 88.57%, and for nouns 89.23%), similar to
the number of correct classifications of real words in
the LCM (84.72%, cf. Semin & Fiedler, 1988).
5
Finally,
participants were instructed to perform a forced choice,
selecting which of the two content domains, agency or
communion, each word matched (with descriptions
adapted from Abele & Wojciszke, 2007; Abele,
Uchronski, Suitner, & Wojciszke, 2008):
A—Agency was introduced as an “orientation toward
actions and being efficient. It is about striving to achieve
goals.”As examples, real words denoting agency were
provided, including two nouns (activity and success),
two adjectives (ambitious and diligent), and two verbs
(strive and achieve).
6
C—Communion was introduced as an “orientation
toward others and focusing on relations with other peo-
ple.”As examples, real words denoting communion
were provided, includingtwo nouns (friendship and po-
liteness), two adjectives (warm and tolerant), and two
verbs (help and support).
The order of the agency and communion descriptions
was counterbalanced. The pseudo-words were
presented in a booklet, with each word presented sepa-
rately on a piece of paper. Besides each word, there
were the two letters A and C, signifying agency and
communion, respectively. The order of the letters
matched the order of the instructions.
Results and Discussion
To determine if the grammatical category predicted
whether participants perceived the word as agentic
(value 1) or communal (value 0), analyses were run
with the grammatical category as a within-participants
5
Crucially, verbs—the focal category for the present research - were
recognized correctly >90% (Studies 2 and 4), significantly more often
than adjectives (Studies 2 and 4) and nouns (Study 2). Misclassifica-
tions predominantly pertained to adjectives and nouns: both, in Studies
2 and 4, most incorrectly recognized adjectives were assigned to the
noun category (86.5% and 85%, respectively). Similarly, most incor-
rectly recognized nouns were assigned to the adjective category (69%
and 78%, respectively). This pattern does not jeopardize the general
claim of this paper that adjectives and nouns are less agentic than verbs
and reflects the fact that in rare cases adjectives have the same suffixas
nouns, even in Polish. As the results of all findings remain robust and
changes in the coefficients are negligible when analyzing only correctly
recognized words—we assume that participants indeed wereguided by
the grammatical categories used in the manipulation.
Table 3. List of stimuli used in all the experiments (GC indicates grammatical category of v—verbs, a—adjectives, n—nouns)
List 1 GC List 2 GC List 3 GC
Stimuli lists for
Studies 2 and 3 (in bold)
nefkiczyćv nefkickie a nefkistwo n
szopfute a szopfustwo n szopfićv
bultestwo n bultewićv bultewne a
łeszdićvłeszdune ałeszdustwo n
bekłowne abekłóstwo nbekłowaćv
kechnystwo n kechnowaćv kechnyckie a
dyżmuwaćvdyżmuwskie a dyżmustwo n
chynfowskie a chynfostwo n chynfowaćv
tyzjastwo ntyzjowaćvtyzjawskie a
fumzićvfumzięte a fumziestwo n
nytczackie a nytczastwo n nytczowaćv
rećwustwo n rećwićvrećwute a
juźbowaćvjuźbuckie a juźbustwo n
bunrowne abunrostwo nbunrićv
fumlystwo nfumlićvfumlyte a
List 1 GC List 2 GC List 3 GC
Stimuli lists for
Study 4
nefkiczyćvnefkidło n (neut) nefkickie a (neut)
tyzja n (fem) tyzjawskie a (neut) tyzjowaćv
dyżmuwskie a (neut) dyżmuwaćvdyżmudłon(neut)
fumzićv fumzyt n (masc) fumzięty a (masc)
szopfudło n (neut) szopfuty a (masc) szopfićv
fumlycki a (masc) fumlićvfumlytn(masc)
juźbowaćvjuźba n (fem) juźbocka a (fem)
chynfyt n (masc) chynfowska a (fem) chynfowaćv
nytczacka a (fem) nytczowaćv nytcza n (fem)
6
Note that in Polish, words are usually clearly categorized into gram-
matical categories (e.g., “the help”and “to help”would be different
words, “pomoc”and “pomagać,”in Polish).
M. Formanowicz et al.Linguistic markers of agency
European Journal of Social Psychology 00 (2017) 00–00 Copyright © 2017 John Wiley & Sons, Ltd.
predictor. Random intercepts were estimated, and two
orthogonal contrasts were created. In the first, verbs
(coded as 2) were compared with the two other cate-
gories (both coded as 1). In the second, adjectives
(1) were compared with nouns (1). The means for
all three experiments are presented in Table 4. A satu-
rated model was estimated in which the two contrasts
were used to predict the agency–communion ratings at
the within-participants level. At the between level,
only the random intercept was estimated. The results
of this model are shown in Table 5. As predicted, the
log odds of choosing agency over communion were
higher for verbs versus the other categories (B=0.16;
SE = 0.06; p<.01; odds ratio = 1.17). There was no dif-
ference between adjectives and nouns in the ascription
of agency versus communion (B=0.05; SE = 0.09;
p= .59; odds ratio= 0.95).
7
When running the analysis
with correct word classifications only, the pattern of
results was confirmed, and the results remain robust.
This pattern first lends support to our claim that
verbs uniquely carry agency information, which is
decoded by perceivers. However, this study has a
number of limitations. First, participants had to
make forced-choice decisions regarding agency–
communion. This dependent variable does not take
into consideration that agency and communion are
in fact two dimensions and that a word could be
perceived as agentic and communal at the same
time. This issue was addressed in the subsequent
studies. Furthermore, prior to making judgments of
agency and communion, participants had to classify
words according to their grammatical category
(adjectives, nouns, verbs, and others), and this task
could have primed them to look for an overlap be-
tween the grammatical classification and the consec-
utive rating task. As this might have biased the
results, the classification task was presented after
the evaluation of the words in the following studies.
Study 3
In Study 3, we assessed agency and communion as sep-
arate dimensions to substantiate the present findings.
Method
Participants.One-hundred and four students par-
ticipated in the study, including 13 men, 58 women,
and 33 people who did not provide information on their
gender (M
Age
=21.03years, SD
Age
= 0.83 years).
Procedure. Given that participants had to rate the
words on several dimensions, we reduced the pseudo-
word sets to nine word stems. Pseudo-words were
selected from the pretest with the same criteria as in
Study 2 and set in a fixed random order. Participants
evaluated paper–pencil lists consisting of nine stimuli
(three verbs, three adjectives, and three nouns). Three
lists were created, each starting with the same word
stem but with a different suffix, and the instructions
matched those provided in Study 2. This time, agency
and communion were assessed using two Likert scales
rather than a forced-choice format (5=opposite of
agency/communion to 5 = perfect example of
agency/communion, cf. Abele & Wojciszke, 2007). As in
Tab l e 4. Means and standard deviations for the dependent variables in Studies 2 to 4 by grammatical category condition
Verbs Adjectives Nouns
MSDMSDMSD
Study 2 Agency/communion 0.61 0.22 0.48 0.23 0.50 0.24
Study 3 Agency 1.44 1.84 0.21 1.62 0.18 1.76
Communion 0.13 1.50 0.10 1.55 0.57 1.72
Abstractness 0.45 2.22 0.60 2.15 0.44 2.06
Valence 0.09 1.57 0.05 1.62 1.14 1.85
Study 4 Agency 0.24 2.07 0.10 1.57 0.17 1.56
Communion 0.37 1.88 0.14 1.60 0.37 1.52
Abstractness 0.26 2.10 0.34 2.10 0.16 2.32
Valence 0.82 1.94 0.08 1.65 0.50 1.79
Note. In Study2, Agency–Communionwas assessed with a forced choice item with communion codedas 0 and agency coded as 1. Higher values indicate a
stronger tendency toward agency choices.
7
An alternative analysis was run in which verbs were contrasted sepa-
rately against adjectives and nouns (with two dummy variables coding
verbs as the reference category). The results were in accordwith the hy-
pothesis. Both adjectives (B=0.53; SE =0.20; p= .007) and nouns
(B=0.43; SE = 0.19; p= .03) were seen as less agentic than verbs.
Tab le 5. Summary of two-level logistic regression analysis predicting
forced choice outcomes in Study 2
Predictor BSEOdds ratio
Within level
Contrast 1 0.16** 0.06 1.17
Contrast 2 0.05 0.09 0.95
Between level
Threshold 0.11* 0.05
Residual variance 0.00 0.00
Note. Contrast 1 contrasts verbs (coded as 2) with adjectives and nouns
(both coded as 1). Contrast 2 contrasts adjectives (1) with nouns (1).
The forced choice dependent variable was coded with communion = 0
and agency = 1.
**p<.01.
*p<.05.
M. Formanowicz et al. Linguistic markers of agency
European Journal of Social Psychology 00 (2017) 00–00 Copyright © 2017 John Wiley & Sons, Ltd.
Study 2, examples of agency and communionwere pro-
vided. Moreover, participants evaluated the valence and
the concreteness/abstractness of words on two scales
(5=negative/concrete to 5 = positive/abstract). The
instructions regarding the latter dimension read as fol-
lows: “Concrete words denote things that exist in reality
and it is easy to picture them; abstract words rather re-
flect thoughts and ideas and do not have physical repre-
sentations.”Examples of concrete words were “shoe,”
“green,”and “to kick,”and examples of abstract words
were “friendship,”“spiritual,”and “to contemplate”
(cf. Brysbaert, Warriner, & Kuperman, 2014, for a sim-
ilar operationalization). Participants received the words
in a booklet. Each word was presented at the top of a
page followed by the evaluation of the four dimensions
in the following order: agency and communion
(counterbalanced), abstractness, and valence. Finally,
for each pseudo-word, participants reported the extent
to which it reminded them of the corresponding real
grammatical category (e.g., verbs for pseudo-verbs) on
ascaleof5, not at all to 5, very much. All three gram-
matical categories were recognized correctly (verbs:
M=2.90, SD = 2.43; adjectives: M=1.72, SD = 2.50; and
nouns: M=2.03,SD = 2.47), a s indicated by a significant
difference from 0, representing the midpoint of the scale
(all ps<.001).
Results and Discussion
Two orthogonal contrasts were created. In the first,
verbs (coded as 2) were compared with the other two
categories (both coded as 1). In the second, adjectives
(1) were compared with nouns (1). The conceptual
model that was tested is presented in Figure 1.
First, we evaluated a saturated model. Second, we
defined additional constraints to test the hypothesis of a
unique verb–agency link. The constraints were as fol-
lows: (i) we set the path from Contrast 2 (comparing ad-
jectives and nouns) to the evaluation of agency to zero, as
we have not hypothesized any difference between these
two categories; and (ii) we set the coefficients for both
contrasts to zero in reference to communion judgments,
as we have not hypothesized any relevance of the gram-
matical categories to the communion dimension.
8
The results of the saturated and the constrained
models, which are congruent with the hypothesized re-
lationships are shown in Table 6. Verbs led to higher
agency perceptions than nouns and adjectives, and
grammatical categories were unrelated to the commu-
nion judgments. When the judgments of the extent to
which artificial words reminded participants of the real
grammatical categories were included in the main anal-
ysis, the overall patternof results was preserved, and the
effect of the similarity to real grammatical categories on
the judgments of agency and communion was negligi-
ble. Together these analyses replicate the results of
Study 2. In extending these studies, we have found that
the perceived abstractness of words was negatively
related to the judgment of agency, meaning that the
more abstractly the word was perceived, the less it was
perceived as agentic.
However, it is noteworthy that the nouns used as
stimuli in this study were biased toward abstractness be-
cause of their suffixes (reserved for abstract words). To
correct for this fact and to test the robustness of our find-
ings, in Study 4 we used suffixes common to broader
classes of nouns (e.g. common nouns with a feminine
vs. masculine vs. neutral grammatical gender).
Study 4
This study was a replication of Study 3 conducted to
substantiate the findings from the previous experi-
ments. New stimuli were used for the grammatical com-
parison group of nouns, and we varied the grammatical
gender of the adjectives.
Method
Participants.One-hundred and twenty-three stu-
dents participated, including 58 men, 65 women
(M
Age
= 20.41 years, SD
Age
=2.44years).
Procedure. There were two major differences
betweenStudies3and4regardingtheexperimental
procedure. The first was the previously described
change in the stimuli (suffixes for common nouns in-
stead of abstract nouns). The second was that partici-
pants classified words in terms of their grammatical
class, as in Study 2 (decision between grammatical cate-
gories instead of continuous ratings), but at the end of
the experiment 83.92% of the words were classified
correctly (for verbs: 93.40%, for adjectives: 74.75%,
and for nouns: 89.82%), as in Study 3.
Results and Discussion
Two orthogonal contrasts were created. In the first,
verbs(codedas2)werecomparedwiththetwoother
categories (both coded as 1). In the second, adjectives
(1) were compared with nouns (1). The conceptual
model that was tested, which is the same as in Study
3, is presented in Figure 1. First, we evaluated the satu-
rated model. Second, we defined additional constraints
in the model to test the hypothesis of a unique verb–
agency link. The constraints were as follows: (i) we set
the path from Contrast 2 (comparing adjectives and
8
An alternative analysis was run with two dummy codes using verbs as
the reference category. The first dummy variable compared adjectives
and the second compared nouns to the verb category. The results were
in accord with the hypothesis. Both adjectives (B=1.19; SE = 0.21;
p<.001) and nouns (B=1.05; SE =0.24;p<.001) were seen as less
agentic than verbs. Compatible with the main analysis, neither adjec-
tives nor nouns differed from verbs in terms of communion (both
ps>.25). When non-significant paths were set to zero and an equality
constraint was placed on the paths leading from the dummy variables
to agency, the model had a very good fit(χ
2
(4) = 2.70; p=.61;
RMSEA = .00; CFI = 1). The results corroborate the hypothesized basic
model and indicate that the verb–agency linkholds equally for compar-
isons with adjectives and nouns.
M. Formanowicz et al.Linguistic markers of agency
European Journal of Social Psychology 00 (2017) 00–00 Copyright © 2017 John Wiley & Sons, Ltd.
Fig. 1: Conceptual model tested in Studies 3 and 4. Contrast 1 contrasts verbs(coded as 2) with adjectives and nouns (both coded as 1). Contrast 2
contrasts adjectives (1) with nouns (1)
Tab l e 6. Estimates and fit indices for saturated and constrained model in Studies 3 and 4
Study 3—saturated Study 3—constrained Study 4—saturated Study 4—constrained
Estimate SE Estimate SE Estimate SE Estimate SE
Within level—agency
Contrast 1 0.37*** 0.07 0.37*** 0.07 0.15* 0.06 0.16** 0.06
Contrast 2 0.07 0.11 0.00 0.00 0.06 0.08 0.00 0.00
Valence 0.17** 0.05 0.16** 0.05 0.26*** 0.04 0.27*** 0.04
Abstractness 0.13*** 0.03 0.13*** 0.03 0.00 0.03 0.003 0.03
Within level—communion
Contrast 1 0.07 0.06 0.00 0.00 0.02 0.53 0.00 0.00
Contrast 2 0.02 0.08 0.00 0.00 0.13 0.07 0.00 0.00
Valence 0.36*** 0.04 0.36*** 0.04 0.40*** 0.03 0.41*** 0.03
Abstractness 0.03 0.03 0.02 0.03 0.05* 0.02 0.05* 0.02
Agency with communion 0.57* 0.25 0.57* 0.25 0.72** 0.22 0.73** 0.22
Residual variances
Agency 4.97*** 0.44 4.97*** 0.44 5.90*** 0.42 5.91*** 0.42
Communion 4.37*** 0.36 4.39*** 0.36 4.64*** 0.35 4.66 0.36
Between level
Agency 0.60*** 0.12 0.60*** 0.12 0.06 0.10 0.06 0.10
Communion 0.19* 0.10 0.19* 0.10 0.22* 0.09 0.22* 0.09
Agency with communion 0.42* 0.19 0.42* 0.19 0.23 0.12 0.22 0.12
Residual variance
Agency 0.89** 0.30 0.89** 0.30 0.49* 0.20 0.49* 0.20
Communion 0.49** 0.16 0.49** 0.16 0.39** 0.13 0.38** 0.13
Model fit
χ
2
(df) 2.16 (3) 3.06 (3)
χ
2
p-value 0.54 0.38
RMSEA 0.0 0.0
CFI 1.0 1.0
TLI 1.0 1.0
SRMR
W
0.01 0.01
SRMR
B
0.002 0.002
Note: Contrast 1 contrasts verbs (coded as 2) withadjectives and nouns (both coded as 1). Contrast 2 contrasts adjectives (1) with nouns (1). Fixed paths
in the constrained model in italics. RMSEA = root mean square error of approximation, CFI = comparative fit index, TLI = Tucker–Lewis index (~NNFI),
SRMR = standardized root mean square residual (W = within, B = between).
***p<.001.
**p<.01.
*p<.05.
M. Formanowicz et al. Linguistic markers of agency
European Journal of Social Psychology 00 (2017) 00–00 Copyright © 2017 John Wiley & Sons, Ltd.
nouns) to the evaluation of agency to zero, as we have
not hypothesized any difference between the two cate-
gories; and (ii) we set the coefficients for both contrasts
to zero in reference to communion judgments, as we
have not hypothesized any relevance of the grammati-
cal categories to the communion dimension. The results
of the saturated and the constrained models are shown
in Table 6. Again, the obtained results are congruent
with the hypothesized relationships. Verbs led to higher
agency ascriptions than nouns and adjectives. More-
over, grammatical categories were unrelated to the
communion judgments.
9
The correlation between
abstractness and agency found in Study 3 was not
evident, because the suffixes of nouns were no longer
confounded with abstractness. All results remained
robust when only including correct classifications in
the analysis.
10
General Discussion
To our knowledge, this is the first set of studies to show
that social judgments are related to meta-semantic char-
acteristics of language. Specifically, the present studies
provide the primary empirical evidence that verbs—
but not adjectives or nouns—are associated with
agency. This verb–agency link is corroborated with evi-
dence from (i) natural language use and (ii) the inter-
pretation of language in controlled experiments. In
actual language use, as investigated with large-scale cor-
pora analyses, stereotypically agentic social targets were
more likely paired with verbs than non-agentic social
targets (Study 1). Importantly, this pattern emerged
for two languages belonging to different language
families, Polish (Slavic) and German (Germanic). More-
over, agency was specifically conveyed by verbs in the
experiments with pseudo-words (Studies 2 through 4),
and this effect was stable even when controlling for
valence and perceived abstraction. The absence of
systematic grammatical category effects on commu-
nion, generally considered the second fundamental
dimension of social judgment, further attests to the
specificity of the verb–agency link. Taken together, this
evidence suggests a strong link between agency and
the grammatical category of verbs, both in language
production and in the construction of meaning from
language.
From a methodological viewpoint, we would like to
highlight that the findings from the corpora analyses
(maximizing ecological validity) and the pseudo-word
studies (maximizing experimental control) converge.
In particular, the pseudo-word paradigm employed
in the three experimental studies has the great advan-
tage that word class and meaning are not
confounded, a problem that limits the validity of
many studies on neural correlates of word classes
(see Vigliocco et al., 2011, for an overview), as well
as most LCM studies.
The verb–agency link fills another blank in the
language-cognition puzzle and may inform future basic
research on the meta-semantic properties of grammatical
categories. Specifically, the relationship of the concrete-
ness of verbs (as implied by the LCM) to agency should
be investigated to consolidate the present approach with
the LCM more concisely. Importantly, the rationale and
results of our studies are complementary to the LCM
account. Our own approach, distinguishing verbs from
other grammatical categories, overlaps only partially with
the more fine-grained LCM model. Within the LCM
framework, SV seem to be a possible exception to the
verb–agency link proposed here. Such verbs mostly refer
to subjects’emotional (and potentially enduring) states
rather than to actions and agency (Brown & Fish, 1983;
Semin, 2000). Hence, they differ from other verb types
mainly on semantic grounds, which goes beyond the
scope of the present research with its focus on meta-
semantic effects. Compared with more common verb
types, SV constitute only a small proportion of verbs,
and, possibly for this reason, Vigliocco et al. (2011) claim
that prototypical verbs refer to actions. Based on learning
theories, it is reasonable to assume that such prototypical,
well-learned associations drive meta-semantic effects. We
used this notion in the pseudo-word studies and assumed
that, when encountering pseudo-verbs, participants
would refer to the central representatives of this gram-
matical category. This notion is supported empirically in
our studies in which participants exposed to pseudo-
words attributed more agency to verbs than to the other
grammatical categories. Thus, they most likely relied on
prototypical associations while ignoring atypical instances
such as SV. However, the interplay between grammatical
categories, agency, and concreteness versus abstractness
should be investigated systematically in the future.
Furthermore, there is potential in investigating the
generalizability of the verb–agency link in other realms.
The likelihood that verbs evoke agentic associations
may be mainly relevant in the social domain regarding
the perception of individuals and groups, because it is
precisely the context of social judgment in which
agency has proven to be an important coordinate in pre-
vious research. Within social judgment, people
described by verbs (or describing themselves in this
way) might be perceived as more agentic than those
addressed with adjectives or nouns. These implications
should be tested in future studies.
9
An alternative analysis was run in which two dummy codes were cre-
ated. The first dummy category compared adjectives and the second
compared nouns with the verb category. The results were in accord
with the hypothesis. Both adjectives (B=0.40; SE = 0.20; p=.04)
and nouns (B=0.52; SE = 0.20; p= .008) were seen as less agentic
than verbs. Compatible with the main analysis, neither adjectives nor
nouns differed from verbs in terms of communion (both ps>.32).
When non-significant paths were set to zero and an equality constraint
was placed on the paths leading from the dummy variables to agency,
the model had a very good fit(χ
2
(4) = 2.41; p= 0.66; RMSEA = .00;
CFI = 1). The results corroborate the hypothesized basic model and in-
dicate that the verb-agency link holds equally for comparisons with ad-
jectives and nouns.
10
All materialsand data sets for Studies 2–4 are available at http://boris.
unibe.ch/83554/
M. Formanowicz et al.Linguistic markers of agency
European Journal of Social Psychology 00 (2017) 00–00 Copyright © 2017 John Wiley & Sons, Ltd.
Outside the social judgment domain, agency may be
less relevant. For instance, when inferring the impor-
tance of an attitude to one’s identity, grammatical cate-
gories may play a role primarily on the basis of their
temporal qualities and related motivations and goals
(cf. Rubini et al., 2014). In fact, in a set of studies on
behavioral effects, people were found to be more
affected by nouns than by verbs when their member-
ships in socially desirable categories were at stake
(Bryan, Adams, & Monin, 2012; Bryan, Walton, Rogers,
& Dweck, 2011). Also, registered voters were shown to
be more likely to vote when questions on voting were in
noun form (“How important is it to you to be a voter in
the upcoming election?”) rather than verb form (“How
important is it to you to vote in the upcoming elec-
tion?”) (Bryan et al., 2011, p. 12653). Similarly, those
remindednottobecheaterswerelesslikelytocheat
than those who were asked not to cheat (Bryan et al.,
2012). Nonetheless, it may still be true that the descrip-
tion “Person X voted”is perceived as more agentic than
the description “Person X was a voter”(i.e., a matter of
social judgment).
However, the comparison of nouns and verbs with
the same word stem may be problematic in this case,
given that the noun is an agent noun formed from
the verb and denotes a person doing this action (e.g.,
the word “eater”is derived from the verb “to eat”). In
the case of “agent nouns,”one could hypothesize that
the agency transfers from the verb to the noun (cf.
Vigliocco et al., 2011). Moreover, such nouns appear
to be used much less frequently than the respective
verbs. In Table 7, we present the corpora frequencies
of a sample of agent nouns and accompanying verbs
that were used in previous studies (Bryan et al., 2011;
Bryan et al., 2012) that attest to this possibility (upper
section of Table 7). As a contrast, we also present a ran-
dom sample of verbs and nouns from the comprehen-
sive list of English lemmas (Brysbaert et al., 2014) in
which a word can be used either as a verb (“to comb”)
or a noun (“a comb”) (lower section of Table 7). In the
former set, nouns consistently evidenced lower fre-
quencies than the associated verbs, which was not true
for the random sample in the second set. This highlights
the uniqueness of “agent nouns.”Encountering them
may make people think that they are used for a reason,
for instance, to highlight the stability of the involved
activity (which would likely trigger more attributions
of agency). Naturally, people try to make sense of their
world and pay close attention to inconsistencies. More-
over, frequency is known to affect the fluency of infor-
mation processing and its consequences (e.g.,
Oppenheimer & Frank, 2008). Therefore, we consider
the frequency issue a possible confound in making
inferences about the grammatical categories of real
words that may also help reconcile the present findings
with previous work.
In the background of these considerations, we had
opted for artificial words for which the concerns about
differences of frequencies of usage or semantics are
not relevant (Studies 2 through 4) as opposed to the
all verbs approach in the corpora analyses (Study 1).
While we recognize that semantics and salience will
guide the social perceiver, we would like to add that
meta-semantic effects, such as the verb–agency link,
may contribute to the sense-making process in a very
subtle way, with the specific venues still to be investi-
gated—a journey we are looking forward to taking.
Conclusion
Verbs express action. This functional property is the
most important feature of the many definitions of this
class of words. For example, according to the Collins
English Dictionary (2003), a verb is “any of a large class
of words in a language that serve to indicate the occur-
rence or performance of an action […].”According to
Random House Kernerman Webster’s College Dictionary
(2010), a verb is defined as “amemberofaclassofwords
that function as the main elements of predicates, typi-
cally express action […].”With this study, we have
shown that this property goes beyond grammar, intrud-
ing into cognition and, by extension, social cognition.
The well-established relationship between language
use and cognition (Semin, 1998) suggests that the lin-
guistic properties shape and constrain cognitive process-
ing of the information conveyed by the linguistic devices
we use, and this has important implications in the social
realm (Holtgraves & Kashima, 2008). For example, pre-
vious studies showed that choosing the first-person pro-
noun moves our attention to the self (Chung &
Pennebaker, 2007), or that dropping the pronoun
moves the attention away from the target performing
the action (Kashima & Kashima, 1998). Interestingly,
in many cases the relationship between language and
cognition is saturated and explained by the semantic
properties of the linguistic devices under scrutiny. In
the previous example, the first-person pronoun (I, me)
semantically expresses the concept of self, and this con-
ceptual priming by the word’s semantics/meaning is
critical for the effect. The LCM (and its extensions) offers
a different perspective that categorizes language accord-
ing to structural properties (at least in the main differ-
ence between adjectives and verbs) as well.
Tab l e 7. Frequency summary of selected agent nouns and infinitives in
three English corpora
Word BYU-BNC COCA GLOWBE
eater / to eat 109 / 2762 714 / 15 987 2953 / 70 790
smoker / to smoke 184 / 265 1292 / 1535 4804 / 4924
runner / to run 666 / 4533 3563 / 23 933 15 699 / 94 914
voter / to vote 279 / 1105 5059 / 9476 29 599 / 57 190
cheater / to cheat 7 / 93 306 / 752 1567 / 4370
a program / to
program
354 / 56 8127 / 597 17 950 / 2373
a trace / to trace 213 / 513 1453 / 1268 2435 / 4651
a rush / to rush 337 / 303 1907 / 1451 4729 / 5590
a taste / to taste 431 / 285 2493 / 5884 9321 / 6707
a whip / to whip 83 / 63 334 / 484 918 / 1961
Note: BYU-BNC = British National Corpus . COCA = Carpus of Contem po-
rary American English. GLOWBE = Corpus of Web-Based Global English.
M. Formanowicz et al. Linguistic markers of agency
European Journal of Social Psychology 00 (2017) 00–00 Copyright © 2017 John Wiley & Sons, Ltd.
However,the concrete implementation of suchstudies
has never fully disentangled the semantic and structural
aspects. For example, being an athlete versus being ath-
letic (Carnaghi et al., 2008, Study 1) is grammaticallydif-
ferent (noun vs. adjective), but it is also semantically
different. Therefore, previous studies typically confirm
thatspecificlanguagedevices(e.g., nouns)areusedto ex-
press specific types of information (e.g.,enduring charac-
teristics), but it was unclear whether this use reflects the
meaning of the word or the very nature of the language
device per se, such that choosing a specificlinguisticde-
vice conveys the corresponding information even in a
context in which previous knowledge cannot contribute
to meaning construction. The pseudo-word paradigm
wasdesignedto testthepurecontributionof the grammar
class, without any further influence embedded in previ-
ous knowledge related to the semantics of the words or
to familiarity because of the frequency of use. Although
the present studies had the main goal of showing that
the grammatical class of verbs is cognitively associated
with agency, they also inform us about a general meta-
semantic effect of grammar. We eventually know about
the meaning of word classes that is independent from
the specific semantic content of the single words. More-
over, we use this knowledge both in natural production
(Study 1) and during information processing when no
otherinformationis present (Studies2 through 4).There-
fore,meaning is not solelyconveyed by semantics; syntax
has a critical role as well. Importantly, the meta-semantic
featureof grammar extendsto the social level. Weappear
to subtly discriminate social targets, and we enforce the
social stereotype whendescribing them throughthe con-
sistent use of grammatical classes by associating stereo-
typically agentic social groups with the grammatical
class that better expresses agency, namely verbs.
In sum, the present studies offer consistent evidence
that basic grammatical categories influencesocialper-
ceptions and that people use these grammatical catego-
ries as a tool for their expressions. The power of
language stems from its pervasiveness and subtlety,
which make it difficult to control, both in usage and
perception. Investigating the verb–agency link within
the language and socialjudgment domains mayhave im-
portant implications for communication in social, legal,
and political domains. Returning to our opening exam-
ple, our results suggest that President Obama was right
in adopting the slogan “Yes we can,”rather than opting
for an adjectival or nominal equivalent. By the same
logic, one may suspect and observe that the recently
founded Spanish party Podemos (Span. “we can”) will be
more successful than the equivalent Italian movement
called Possibile (Ital. “possible”). Our results suggest that
Obama’s verb slogan and the party labelPodemos convey
the agency required to introduce the proposed change.
Conflict of Interest
The authors declare that there are no potential conflicts
of interest with respect to the research, authorship,
and/or publication of this article.
Acknowledgements
This research was supported by the grant WP/BST/IND/
2012/B/8 awarded to the first author from the Polish
Ministry of Science and Higher Education as core
funding for statutory research for the SWPS University
of Social Sciences and Humanities, Faculty of
Psychology.
References
Abele,A.E.,Cuddy,A.J.C.,Judd,C.M.,&Yzerbyt,V.Y.
(2008). Fundamental dimensions of social judgment.
European Journal of Social Psychology,38,1063–1065.
http://dx.doi.org/10.1002/ejsp.574
Abele, A. E., Rupprecht, T., & Wojciszke, B. (2008). The influ-
ence of success and failure experiences on agency.
European Journal of Social Psychology,38,436–448. http://
dx.doi.org/10.1002/ejsp.454.
Abele, A. E., Uchronski, M., Suitner, C., & Wojciszke, B.
(2008). Towards an operationalization of the fundamen-
tal dimensions of agency and communion: Trait content
ratings in five countries considering valence and fre-
quency of word occurrence. European Journal of Social
Psychology,38,1202–1217. http://dx.doi.org/10.1002/
ejsp.575
Abele, A. E., & Wojciszke, B. (2007). Agency and communion
from the perspective of self versusothers. Journal of Person-
ality and Social Psychology,93,751–763. http://dx.doi.org/
10.1037/0022-3514.93.5.75
Abele, A. E., & Wojciszke, B. (2014). Communal and agentic
content in social cognition: A dual perspective model. J. M.
Olson, & M. P. Zanna (Eds.), Advances in experimental social
psychology (pp 195–255). San Diego, CA: Academic Press.
Bettinsoli, M. L., Maass, B., Kashima, Y., & Suitner, C. (2015).
Word-order and causal inference: The temporal attribution
bias. JournalofExperimentalSocialPsychology,60,144–149.
http://dx.doi.org/10.1016/j.jesp.2015.05.011
Brown, R., & Fish, D. (1983). The psychological causality
implicit in language. Cognition,14,237–273. http://dx.doi.
org/10.1016/0010-0277(83)90006-9
Bruckmüller, S., & Abele, A. E. (2013). The density of the big
two: How are agency and communion structurally repre-
sented? Social Psychology,44,63–74. http://dx.doi.org/
10.1027/1864-9335/a000145
Bryan, C. J., Adams, G. S., & Monin, B. (2012). When
cheating would make you a cheater: Implicating the self
prevents unethical behavior. Journal of Experimental Psy-
chology: General.Advance online publication. http://dx.doi.
org/10.1037/a0030655
Bryan, C. J., Walton, G. M., Rogers, T., & Dweck, C.S. (2011).
Motivating voter turnout by invoking the self. Proceedings of
the National Academy of Sciences of the United States of America,
108,12653–12656. http://dx.doi.org/10.1073/pnas.
1103343108
Brysbaert, M., Warriner, A. B., & Kuperman, V. (2014).
Concreteness ratings for 40 thousand generally known
English word lemmas. Behavior Research Methods,46,
904–911. http://dx.doi.org/10.3758/s13428-013-0403-5
Caramazza, A., & Hillis, A. E. (1991). Lexical organization of
nouns and verbs in the brain. Nature,349,788–790.
http://dx.doi.org/10.1038/349788a0
M. Formanowicz et al.Linguistic markers of agency
European Journal of Social Psychology 00 (2017) 00–00 Copyright © 2017 John Wiley & Sons, Ltd.
Carnaghi, A., Maass, A., Gresta, S., Bianchi, M., Cadinu, M.,
& Arcuri, L. (2008). Nomina sunt omina: On the inductive
potential of nouns and adjectives in person perception.
Journal of Personality and Social Psychology,94,839–859.
http://dx.doi.org/10.1037/0022-3514.94.5.83
Chung, C., & Pennebaker, J. W. (2007). The psychological
functions of function words. In K. Fiedler (Ed.), Social com-
munication (pp. 343–359). New York, NY: Psychology
Press.
Coenen, L., Hedebouw, L., & Semin, G. R (2006). Measuring
abstraction: The linguistic category model. Retrieved from
http://cratylus.org/resources/linguistic-model
Collins English dictionary—Complete and unabridged.
(2003). Retrieved September 03 2015 from http://www.
thefreedictionary.com/Verbs
Cuddy, A. J. C., Fiske, S. T., & Glick, P. (2008). Warmth and
competence as universal dimensions of social perception:
The stereotype content model and the BIAS map. Advances
in Experimental Social Psychology,40,61–149. http://dx.doi.
org/10.1016/S0065-2601(07)00002-0
Diekman, A. B., & Eagly, A. H. (2000). Stereotypes as dynamic
constructs: Women and men of the past, present, and future.
Personality and Social Psychology Bulletin,26,1171–1188.
http://dx.doi.org/10.1177/0146167200262001
Eagly, A. H., & Karau, S. J. (2002). Role congruity theory of
prejudice toward female leaders. Psychological Review,109,
573–598 http://dx.doi.org/10.1037//0033-295X.109.3.57
Fausey, C. M., & Boroditsky, L. (2010). Subtle linguistic cues
influence perceived blame and financial liability.
Psychonomic Bulletin & Review,17,644–650. http://dx.doi.
org/10.3758/PBR.17.5.644
Fausey, C. M., Long, B. L., Inamori, A., & Boroditsky, L.
(2010). Constructing agency: The role of language. Fron-
tiers in Psychology,1(162), 1–11. http://dx.doi.org/
10.3389/fpsyg.2010.0016
Fiedler, K. (2008). Language: A toolbox for sharing and
influencing social reality. Perspectives on Psychological Science,
3,38–47. http://dx.doi.org/10.1111/j.1745-6916.2008.
00060.x
Fiske, S. T., Cuddy, A. J., Glick, P., & Xu, J. (2002). A model of
(often mixed) stereotype content: Competence and
warmth respectively follow from perceived status and
competition. Journal of Personality and Social Psychology,82,
878–902. http://dx.doi.org/10.1037//0022-3514.82.6.87
Foroni, F., & Semin, G. R. (2009). Language that puts you
in touch with your bodily feelings: The multimodal
responsiveness of affective expressions. Psychological
Science,20,974–980. http://dx.doi.org/10.1111/j.1467-9
280.2009.02400
Gelman, A., & Heyman, D. (1999). Carrot-eaters and
creature-believers: The effects of lexicalization on children’s
inferences about social categories. Psychological Science,10,
489–493. http://dx.doi.org/10.1111/1467-9280.0019
Holtgraves, T. M., & Kashima, Y. (2008). Language, meaning,
and social cognition. Personality and Social Psychology Review,
12,73–94. http://dx.doi.org/10.1177/1088868307309605
Kashima, E. S., & Kashima, Y. (1998). Culture and language:
The case of cultural dimensions and personal pronoun use.
Journal of Cross-Cultural Psychology,29,461–486. http://dx.
doi.org/10.1177/0022022198293005
Kenworthy, J. B., & Tausch, N. (2008). Expectations about
the accuracy and stability of warmth versus competence
traits: An intergroup analysis. European Journal of Social Psy-
chology,38,1121–1129. http://dx.doi.org/10.1002/
ejsp.543
Kroeger, P. (2005). Analyzing grammar: An introduction.
Cambridge, UK: Cambridge University Press.
Kupietz, M., Belica, C., Keibel, H., & Witt, A. (2010). The
German Reference Corpus DeReKo: A primordial sample
for linguistic research. In: Calzolari, Nicoletta et al. (Eds.):
Proceedings of the Seventh Conference on International Language
Resources and Evaluation (LREC 2010, pp. 1848–1854). Re-
trieved from: http://www.lrec-conf.org/proceedings/
lrec2010/pdf/414_Paper.pdf
Muthén, L., & Muthén, B. (1998-2012). Mplus user’s guide.
Los Angeles, CA: Muthén & Muthén. Available from
http://www.statmodel.com.
Oppenheimer, D. M., & Frank, M. C. (2008). A rose in any
other font would not smell as sweet: Effects of perceptual
fluency on categorization. Cognition,106,1178–1194.
http://dx.doi.org/10.1016/j.cognition.2007.05.010
Pęzik, P. (2012). Wyszukiwarka PELCRA dla danych NKJP.
Narodowy Korpus Języka Polskiego [PELCRA Search En-
gine for Polish National Corpus]. (Przepiórkowski A.,
Bańko M., Górski R., Lewandowska-Tomaszczyk B.
(red.). Wydawnictwo PWN.
Random House Kernerman Webster’s college dictionary.
(2010). Retrieved September 03 2015 from http://www.
thefreedictionary.com/Verbs
Rubini, M., Menegatti, M., & Moscatelli, S. (2014). The
strategic role of language abstraction in achieving sym-
bolic and practical goals. European Review of Social Psychol-
ogy,25,263–313. http://dx.doi.org/10.1080/10463283.20
14.985501
Semin, G. R. (1998). Cognition, language, and communica-
tion. In S. R. Fussel, & R. J. Kreuz (Eds.), Social and cognitive
approaches to interpersonal communication (pp. 229–257).
Hillsdale, NJ: Erlbaum.
Semin, G. R. (2000). Language as a cognitive and behavioral
structuring resource: Question—answer exchanges.
European Review of Social Psychology,11(1), 75–104. http://
dx.doi.org/10.1080/14792772043000004
Semin, G. R., & Fiedler, K. (1988). The cognitive functions of
linguistic categories in describing persons: Social cognition
and language. Journal ofPersonality and Social Psychology,54,
558–568. http://dx.doi.org/10.1037/0022-3514.54.4.55
Semin, G. R., & Fiedler, K. (1991). The linguistic category
model, its bases, applications and range. In W. Stroebe, &
M. Hewstone (Eds.), European review of social psychology
(pp. 1–30), 2. New York, NY: Wiley.
Semin, G. R., & Marsman, J. G. (1994). Multiple inference-
inviting properties of interpersonal verbs: Event instiga-
tion, dispositional inference, and implicit causality. Journal
of Personality and Social Psychology,67,836–849. http://dx.
doi.org/10.1037/0022-3514.67.5.836
Śledziński, D. (2010). Fonemy, difony, trifony i sylaby—
charakterystyka jednostek na podstawie korpusu
tekstowego [Phonemes, diphones, triphones and syllables
—unit characteristics based on textual corpora]. Kwartalnik
Językoznawczy,2010,3–4.
Suitner, C., & Maass, A. (2008). The role of valence in the
perception of agency and communion. European Journal
of Social Psychology,38,1073–1082. http://dx.doi.org/
10.1002/ejsp.52
M. Formanowicz et al. Linguistic markers of agency
European Journal of Social Psychology 00 (2017) 00–00 Copyright © 2017 John Wiley & Sons, Ltd.
Twenge, J. M. (1997). Changes in masculine and feminine
traits over time: A meta-analysis. Sex Roles,36,305–325.
http://dx.doi.org/10.1007/BF02766650
Uchronski, M., Abele, A. E., & Bruckmüller, S. (2013).
Empathetic perspective taking and the situational mal-
leability of the communal self-concept: How
perspective-taking affects self-descriptions. Self and Iden-
tity,12,236–256.
Vigliocco, G., Vinson, D. P., Druks, J., Barber, H., & Cappa,
S. F. (2011). Nouns and verbs in the brain: A review of
behavioural, electrophysiological, neuropsychological
and imaging studies. Neuroscience and Biobehavioral
Reviews,35,407–426. http://dx.doi.org/10.1016/j.
neubiorev.2010.04.00
Walton, G. M., & Banaji, M. R. (2004). Being what you say:
The effect of essentialist linguistic labels on preferences.
Social Cognition,22,193–213. http://dx.doi.org/10.1521/
soco.22.2.193.35463
Wigboldus, D. H. J., & Douglas, K. M. (2007). Language,
stereotypes, and intergroup relations. In K. Fiedler (Ed.),
Social communication (pp. 79–106). New York, NY: Psychol-
ogy Press.
Willems, R. M., Hagoort, P., & Casasanto, D. (2010). Body-
specific representations of action verbs: Neural evidence
from right- and left-handers. Psychological Science,21,
67–74. http://dx.doi.org/10.1177/095679760935407
Ybarra, O., Chan, E., Park, H., Burnstein, E., Monin, B., &
Stanik, C. (2008). Life’s recurring challenges and the fun-
damental dimensions: An integration and its implications
for cultural differences and similarities. European Journal
of Social Psychology,38,1083–1092. http://dx.doi.org/
10.1002/ejsp.55
M. Formanowicz et al.Linguistic markers of agency
European Journal of Social Psychology 00 (2017) 00–00 Copyright © 2017 John Wiley & Sons, Ltd.
