Real and Fictive Motion Processing

in Polish L2 Users of English and

Monolinguals: Evidence for Different
Conceptual Representations
EWA TOMCZAK ANNA EWERT
Adam Mickiewicz University in Poznań Adam Mickiewicz University in Poznań
Faculty of English Faculty of English
al. Niepodległości 4 al. Niepodległości 4
61-874, Poznań 61-874, Poznań
Poland Poland
Email: [email protected] Email: [email protected]

We examine cross-linguistic influence in the processing of motion sentences by L2 users from an

embodied cognition perspective. The experiment employs a priming paradigm to test two hypotheses
based on previous action and motion research in cognitive psychology. The first hypothesis maintains
that conceptual representations of motion are embodied by sensory and motor systems (Glenberg &
Kaschak, 2002; Zwaan & Taylor, 2006) and predicts that linguistic descriptions of motion will be
susceptible to conceptual priming effects. The second hypothesis assumes that the processing of fictive
motion sentences, such as The road runs through the forest (cf. Talmy, 2000a), involves additional mental
simulation of motion (Matlock, 2004b; Richardson & Matlock, 2007) and predicts that fictive motion will
be processed more slowly than real motion. The participants were groups of Polish and English
monolinguals and Polish L2 users of English, who were students in an English-medium university
program. Sentence meaningfulness judgment and response time data were collected from the mono-
linguals and the L2 users tested in both languages. Sentence meaningfulness judgments were examined
to see how the participants represent motion in language and response time data were analyzed for
evidence of conceptual processing. The results showed expected differences in response times to various
sentence types in all groups, but the priming effect was not observed in Polish monolinguals and only
selectively present in the L2 users responding in both languages. The results of our experiment indicate
that fictive motion processing is both embodied in physical experience of motion and modified by
language.
Keywords: motion verbs; fictive motion; L1/L2 processing; embodied cognition; conceptual processing

THE PRESENT STUDY EXAMINES COGNITIVE the advanced level. The conceptual domain of
representations and online processing of real motion is studied because the experience of
(factive) and fictive motion descriptions in motion is universal; motion pervades all aspects
monolinguals and advanced second language of human life, and motion-related language
(L2) users. It assumes that differences in repre- permeates to those areas of linguistic expression
sentation and processing between monolingual where no real movement is described. Within the
speakers and L2 speakers are best investigated at extensive literature on transfer in the domain of
motion, this is the first study investigating fictive
The Modern Language Journal, 99, Supplement, (2015) motion, that is, “motion with no physical occur-
DOI: 10.1111/modl.12178 rence” (Talmy, 2000a, p. 99). The present study is
0026-7902/15/49–65 $1.50/0 also distinct in its choice of theoretical framework:
© 2015 The Modern Language Journal Instead of relying on Talmy’s (1985, 2000b)
50 The Modern Language Journal 99 (2015)

typology of lexicalization patterns, Slobin’s (e.g., attention to a particular characteristic of motion

1996, 2006) Thinking for Speaking approach, or might have some cognitive consequences.
drawing on the linguistic relativity tradition, it Talmy (1985, 2000b) divides languages into
investigates figurative motion from an embodied satellite-framed and verb-framed depending on
cognition perspective. how they encode path of motion. In satellite-
We begin by discussing theoretical perspectives framed languages, such as English or Polish,
and major findings in second language acquisi- verbs convey primarily manner of motion, while
tion (SLA) motion research as well as typological path of motion is encoded in a satellite, that is, a
differences. We then present the embodied preposition or a verbal particle (for a critical
cognition perspective along with its insights discussion, see Pavlenko & Volynsky, 2015). In
regarding action and motion. On that basis, verb-framed languages, such as French or Span-
we formulate two hypotheses—the grounding ish, the main verb in a sentence conveys path of
hypothesis and the simulation hypothesis—which motion, while manner is optional, expressed by
we test experimentally with Polish L2 users of adverbial phrases or other verbs. Talmy refers to
English as well as two monolingual groups these relationships between meaning and form
(English and Polish) using a conceptual priming as lexicalization patterns. They have also been
paradigm. The experiment yields two types of referred to as conceptualization patterns (e.g.,
data that provide information on participants’ Bylund & Jarvis, 2011; Flecken, 2011) to empha-
conceptual processing (cf. Simmons et al., 2008): size that the processes of constructing pre-verbal
(a) meaningfulness judgments and (b) response messages—in terms of selection and linearization
times. We hypothesize that participants will of information to be encoded in language—take
process motion sentences faster if action verb place in what Levelt (1989) calls the concep-
primes indeed activate mental representations of tualizer and extend beyond lexico–semantic
movement and directionality. relations. Additionally, conceptualizations of
motion in different languages are affected by
COGNITION, CONCEPTUALIZATION features such as perspective taking or grammati-
PATTERNS, AND TYPOLOGICAL cal aspect (Bylund & Jarvis, 2011; von Stutter-
DIFFERENCES heim, 2003).
Research conducted in this tradition with L2
The domain of motion has so far been users of typologically different languages dem-
investigated in L2 research from the linguistic onstrates various phenomena: forward transfer of
relativity (Whorf 1940/1956) and the Thinking- conceptualizations from L1 to L2 (e.g., Schmied-
for-Speaking perspective (Slobin, 1996, 2006). In tová, von Stutterheim, & Carroll, 2011), reverse
principle, the Whorfian hypothesis assumes transfer from the L2 to the L1 (Brown &
different cognitive representations of events in Gullberg, 2010; Bylund & Jarvis, 2011); bidirec-
speakers of different languages and is frequently tional transfer (Hohenstein, Eisenberg, &
operationalized as the influence of linguistic Naigles, 2006), a bilingualism effect whereby L2
categorizations on participants’ behavior in users’ conceptualizations differ from the con-
nonlinguistic tasks (e.g., Athanasopoulos & By- ceptualizations of monolingual speakers of both
lund, 2013; Thierry et al., 2009). On the other their L1 and L2 (Brown & Gullberg, 2008;
hand, Slobin (1996) assumes that Hohenstein et al., 2006), and an influence of
age of acquisition (Bylund, 2009; Hohenstein
the grammaticized categories that are most suscepti- et al., 2006). The focus of these studies has been
ble to SL [source language] influence (...) cannot be on how speakers of different languages describe
experienced directly in our perceptual, sensory– motion events verbally (e.g., Bylund, 2009;
motor, and practical dealings with the world. To be
Pavlenko, 2010; Schmiedtová et al., 2011;
sure, all human beings experience sequences of
events that have particular temporal contours, put
von Stutterheim, 2003), what they communi-
objects in locations, and so on. (p. 91) cate in gesture (Brown & Gullberg, 2008), or
where they look when describing motion event
scenes visually presented to them (e.g.,
Since languages employ different grammatical Flecken, 2011). Nonverbal experiments in the
and lexical means to describe the same motion Whorfian tradition, tapping into the percep-
events, different information structures will be tion of motion scenes with no overt linguistic
required to describe perceptually identical events. task, have produced mixed results (e.g., Athana-
In his more recent writings, Slobin (e.g., 2006) sopoulos & Bylund, 2013; Czechowska & Ewert,
acknowledges that language-driven repeated 2011).
Ewa Tomczak and Anna Ewert 51

EMBODIMENT sentences—sentences that contain reference to

motion but do not imply physical movement—do
A hypothesis that so far has not been exploited not activate primary motor cortices but do engage
in cross-linguistic or SLA research on motion is areas involved in higher order motor processes.
the embodiment hypothesis, which maintains Finally, idiomatic action sentences—figurative sen-
that linguistic representations of motion are tences that contain an action verb but the
grounded in sensory–motor experiences of mo- meaning of the whole sentence does not imply
tion (e.g., Barsalou, 1999, 2008; Wilson, 2002). motion of any kind—do not engage motor areas
This view is supported by behavioral (e.g., at all.
Glenberg & Kaschak, 2002; Zwaan & Taylor, Taken together, this means that individuals’
2006) as well as neuroimaging evidence (e.g., conceptual representations of motion language
Pulvermüller, Härle, & Hummel, 2001; Tetta- have perceptual and sensory counterparts in the
manti et al., 2005), demonstrating that sensory brain. Since different languages structure the
and motor cortices are co-activated while motion information about motion events differently, it is
language is being processed by the brain. These possible that speakers of different languages will
co-activation patterns are frequently explained as process descriptions of different kinds of motion
mental simulation of physical action that is taking differently. Based on previous research in cogni-
place while linguistic concepts of motion are tive psychology, we designed an experiment to
being evoked. It is tacitly assumed that this kind of test this prediction.
reenactment of motion is universal, except that it Since the embodiment perspective in L2
is modulated by individual experiences of action: motion cognition has recently emerged (e.g.,
For example, ice hockey players will have different Bergen et al., 2010; De Grauwe et al., 2014), it is
representation of action and different activations necessary to probe more deeply into its assump-
from nonplayers. tions and claims. Thus far, it has been tacitly
Recently, this version of embodied cognition assumed that the results obtained with speakers of
has been questioned with the argument that one language apply as well to speakers of other
the interactions among language, perception, languages. However, inasmuch as most of this
and action in the brain are more complex research has been conducted with monolingual
and context-dependent than most simulation speakers of English, it remains to be seen whether
accounts would have it. Thus, while “language, speakers of other languages, including L2 speak-
perception, and action are not isolated modules, ers of those languages, ground motion in the
but rather interact dynamically” (Bedny & Car- same way.
amazza, 2011, p. 92), it is highly unlikely that the
brain faithfully reenacts physical movement while REAL AND FICTIVE MOTION
processing motion language, since the patterns of
brain activations do not correspond to the actual Real motion, encoded in sentences like John
limb movements required. Accordingly, Chatter- runs through the forest, is referred to by Talmy
jee (2010) proposes a graded grounding pers- (2000a) as factive motion. The term factive implies
pective, in which conceptual representation, truth and an overlap with reality, thus factive
including motor and spatial information, is motion is used with reference to a concrete, overt
distributed across the brain in a way that involves (explicit) movement of some entity through
both motoric and abstract properties of action. In a three-dimensional space. In contrast, fictive
a similar vein, more recent imaging research motion, encoded in sentences like The road runs
(Desai et al., 2013) suggests that semantic through the forest is defined by Talmy as “motion
processing of action language is context sensitive with no physical occurrence” (p. 99), that is, it
and different types of motion sentences involve refers to an event in which no real movement
the motor cortices of the brain in different ways. takes place. In fictive motion “a factive represen-
This means that the linguistic context in which a tation of an object as stationary is held in
motion verb is embedded modulates the activa- cognition concurrently with a fictive representa-
tion of sensory–motor areas of the brain. Specifi- tion of the same object as moving” (Talmy, 2013,
cally, abstract sentences—sentences that contain p. 23). As Talmy underscores, the distinction
no reference to action—do not activate motor between factive and fictive by no means attempts
cortices in the brain. Literal action sentences— to highlight that a given cognitive representation
sentences describing real motion—activate pri- is objectively real or unreal; instead, it embraces
mary motor cortices as well as areas implicated in the degree of veridicality (that is, the degree of
planning physical movement. Figurative action likelihood) ascribed to a mental representation of
52 The Modern Language Journal 99 (2015)

a given entity by one of our cognitive subsystems. deaf users of British Sign Language (BSL). They
Viewed through such a prism, factive is more replicated Glenberg and Kaschak’s (2002) study
veridical (e.g., A man jogs along the Pacific coastline), with close translations of their English directional
and fictive is less veridical (e.g., The Andes run sentences into BSL. They found no ACE effect in
along the Pacific Coast of South America). BSL sentence comprehension across their experi-
In cognitive linguistic terms, all motion events, ments, but reported ACE effects in the same
both factive and fictive, have a corresponding population in written language comprehension
structure. Following Talmy (1985), we assume (in their replication of the original Glenberg &
that a “[m]otion event consists of one object (the Kaschak [2002] study). Their findings suggest
Figure) moving or located with respect to another that viewing sign language does not engage action
object (the reference object, or Ground). It is simulation in the way that reading written
analyzed as having four components: besides language or listening to spoken language does,
Figure and Ground, there are Path and Motion” as perceiving hand movements in sign language
(p. 61). It should be noted that much research in may block the involvement of the motor system in
cognitive psychology relates to movement but not language comprehension.
motion as defined by Talmy. We refer to those Research integrating co-verbal gesture produc-
instances as action rather than motion, although tion from an embodied cognition perspective
the postulated processing mechanisms for both provides more insight. Hostetter and Alibali
are the same. (2008) presented the gesture-as-simulated-action
(GSA) framework in which “gestures emerge
Real Motion and the Grounding Hypothesis from the perceptual and motor simulations that
underlie embodied language and mental imag-
What we shall call the grounding hypothesis is the ery” (p. 502).
embodied cognition assumption that action and As presented earlier, the congruent facilitation
motion language processing is grounded in effect has been interpreted as evidence that
bodily activity (Barsalou, 1999, 2008; Wilson, language processing recruits motor processes.
2002). This view holds that what we know about Since verb primes and verb phrases have been
the surrounding world is deeply rooted in our shown to facilitate the execution of prime-related
perceptual representations that have been built movement (a nonlinguistic stimulus), we hypoth-
up through our interactions with objects and esize that priming effects should also be present
performing actions by the self. during the processing of linguistic targets. Along
Numerous behavioral studies on motion proc- with the facilitation view, literal action sentences
essing demonstrate that observing an action that as well as fictive motion sentences are predicted to
is either explicit (shown in a video) or more be processed faster when preceded by a verb
implicit (suggested by picture content) has a prime than when preceded by a noun prime.
facilitating effect on action (Tucker & Ellis, 2004;
Zwaan & Taylor, 2006). Such an effect is Fictive Motion
frequently referred to as an action priming effect.
Experiments quite consistently show that reac- Fictive motion expressions, present cross-lin-
tion time measures obtained for the targets are guistically (e.g., in English, Hindi, Japanese,
shorter when these are preceded by primes Polish, Spanish), rely on the lexicalization pat-
congruent in terms of the nature and direction terns and semantic components (cf. Talmy,
of motion. This observation has been reported in 2000a) employed by factive motion events, where
the studies that employed visuo–motor as well as some figure of motion is presented as traversing
motor–visual priming paradigms. What is meant some ground along some path of motion and
by visual is either related to watching a video (or a in some manner of motion. Fictive motion expre-
picture where movement is implicit), or to ssions are figurative as they feature a figure that
reading action verbs or action sentences; what is inanimate and incapable of self-movement, yet
is meant by motor is the very enactment of a is presented as dynamic.
movement. Talmy (2000a) argues that humans are able to
Remarkably, behavioral research suggests that understand sentences featuring an inanimate
the engagement of action simulation in language figure as moving through space thanks to “the
comprehension may be dependent on the imaginal capacity of cognition” (p. 100), or
modality of language presentation. In a series of human ability to think in abstract terms. Fictive
studies, Perniss et al. (2013) investigated the motion is believed to be subjectively experienced
Action–Sentence Compatibility Effects (ACE) in by the so-called conceptualizer (a speaker or
Ewa Tomczak and Anna Ewert 53

listener) by mentally enacting motion in the sentences. Hence, fictive motion can be predicted
absence of an explicit agent (Matlock, 2004b). to take more time to process than motion that
This subjective imagery involves an implicit is expressed nonfiguratively. Since simulation
mental simulation of movement through a accounts have been frequently provided to
three-dimensional space. explain activations in sensory and motor areas
of the brain in the processing of motion in
Coextension Paths. The first comprehensive
general, for reasons of clarity, from this point on
taxonomy of linguistic manifestations of fictive
we will refer to the simulation hypothesis in fictive
motion (FM) was proposed by Talmy (2000a),
motion processing exclusively.
who identified emanation paths, pattern paths,
The support for the simulation view comes
frame-relative motion, advent paths, access paths,
from priming experiments in the field. Matlock’s
and coextension paths. Recent theoretical and
(2004a) experimental results suggest that, in
experimental investigations have chiefly focused
order to understand a FM sentence, “people
on coextension paths (e.g., The fence goes/zigzags/
re-activate and simulate aspects of the pro-
descends from the plateau to the valley) (cf. Talmy,
tagonist’s motion, including speed, distance,
2000a) and their subtypes. The very need for
and the terrain across which the movement
making a fine-grained distinction within coexten-
occurred. In doing so, they construct a dynamic
sion paths was independently emphasized by
representation that mirrors the actual motion of
Matsumoto (1996) and Matlock (2004a). Matsu-
the protagonist” (p. 237). In Matlock’s (2004b)
moto noticed that low-manner-salient languages
studies, participants first read a story about travel
(e.g., Japanese, Spanish) tend to have a somewhat
that was either fast or slow, short- or long-
more restricted choice of figures available for
distanced, easy or difficult, and then made a
constructing fictive motion sentences than high-
timed decision about a fictive motion sentence
manner-salient languages (e.g., English, Polish).
that followed. The priming condition with the
In fact, in some languages there is a difference in
description of difficult travel yielded longer RTs
treating linear figures that are travelable (e.g.,
than the condition where the journey was
paths, roads, etc.) and figures that are also linear
described as easy or fast. There is also evidence
but are not intended for human travel, hence
from priming experiments employing picture
nontravelable (e.g., walls, fences, and ropes). For
drawing and arrow drawing that supports the
instance, in Japanese some nontravelable linear
claim that people really activate motion informa-
entities such as fences and walls are not amenable
tion while reading and conceptualizing static
to fictive motion description, whereas other, such
scenes that are verbally depicted by figurative uses
as wires or borders, participate in FM construc-
of motion verbs (Matlock, 2006).
tions, yet allow for a narrower repertoire of
Strong support for simulating motion while
motion verbs that can be coupled with them (cf.
processing fictive motion sentences comes from
Matsumoto, 1996). A comparable constraint on
eye-tracking studies. Records of eye movements,
figure participation in fictive motion has been
considered to be strongly related to mental
reported for Spanish (cf. Rojo & Valenzuela,
simulation, show that participants produce longer
2009).
gaze durations and more fixations in response to
Matsumoto’s enriched taxonomy of coextension
FM sentences than while processing static depic-
paths overlaps with the typology proposed by
tions (cf. Matlock & Richardson, 2004; Mishra
Matlock (2004a) who points to a necessary
& Singh, 2010; Richardson & Matlock, 2007).
distinction between expressions that incorporate
However, it seems that motion simulations can be
(a) trajectors (figures) that are ordinarily associ-
modulated by morphosyntax, as research with
ated with motion and that can be coupled with
speakers of Hindi shows that gaze durations in
manner verbs (e.g., The highway crawls through the
comprehension of fictive motion sentences might
city), and (b) paths that are not normally
be modified by word order and, what is more,
associated with motion (e.g., The table goes from
Hindi participants deploy the highest visual
the kitchen to the sliding door).
attention during the comprehension of the
Simulation Hypothesis. It is frequently hypothe- noun while processing fictive motion sentences
sized that the processing of fictive motion (Mishra & Singh, 2010).
sentences involves additional mental simulation
of motion (Matlock, 2004b; Matlock & Richard- HYPOTHESES
son, 2004; Mishra & Singh, 2010; Richardson
& Matlock, 2007) that, overall, lengthens the On the basis of previous research on motion
processing time needed for understanding such processing, we have formulated two hypotheses.
54 The Modern Language Journal 99 (2015)

The grounding hypothesis assumes that conceptual range 21–23). The English monolinguals were
representations of motion are embodied by native speakers of American English tested in the
sensory and motor systems and predicts that United States. The Polish monolinguals and L2
linguistic descriptions of motion, both real and users of English were students at Adam Mickiewicz
fictive, will be susceptible to conceptual priming University in Poznań. The participants completed
effects. By contrast, static depictions in which no the Language Experience and Proficiency Ques-
motion verb appears will not be affected by the tionnaire (LEAP–Q) (Marian, Blumenfeld, &
prime. Different response times in different Kaushanskaya, 2007) that provided information
languages under different priming conditions on their levels of proficiency (speaking, under-
would mean that different languages ground standing spoken language, and reading) in all
motion differently, possibly because of their languages they knew, along with language expo-
morphological or morphosyntactic peculiarities. sure, language history, frequency of use, and
The second hypothesis, the simulation hypothesis, cultural affiliation. The analyses of LEAP–Q self-
assumes that the processing of fictive motion reported data revealed that all three groups were
involves simulation of physical motion and homogenous and comparable in terms of age and
predicts that fictive motion will be processed educational level. Self-reported proficiency in all
more slowly than static depictions and real the groups was similar, with the L2 users rating
motion. In what follows, these hypotheses will themselves somewhat lower in the L2. Mean self-
be tested experimentally. reported proficiency is provided in Table 1.
Target Stimuli. 128 target sentences (64 critical
sentences and 64 filler sentences) were used in
EXPERIMENT
each language version of the experiment (English
The experiment is a conceptual priming task and Polish). The stimuli of interest included 64
with sentences describing two types of fictive sentences of four different types: fictive motion with
motion entities (travelable and nontravelable), travelable entities (FM Tr), fictive motion with
real motion, and static depictions. An unmasked nontravelable entities (FM NTr), real motion (RM),
conceptual priming paradigm was employed that and static depictions (ST). For each sentence type,
involved a sentence meaningfulness judgment 16 short sentences were tested. Each tested
task. Participants’ response latencies and sen- sentence had counterparts across all four sen-
tence meaningfulness judgments were recorded tence types. An example of the between-categori-
with a timing resolution of 1 millisecond using the cal sentence correspondence follows. The English
DirectRT v.2008/2010 software. All participants sentences had their translation equivalents in
went through the same experimental session Polish.
procedures in their respective languages.
EXAMPLE 1. Between-Categorical Sentence Corre-
spondence
Participants
English:
Sixty participants volunteered to take part in a. The footpath crawls between the trees. (FM Tr)
the experiment. Based on their linguistic profile, b. The ditch crawls between the trees. (FM NTr)
c. John crawls between the trees. (RM)
the participants formed three groups: English
d. The footpath is between the trees. (ST)
monolinguals (n ¼ 20; 11 female, 9 male; mean Polish:
age ¼ 20.65; age range 19–26), Polish monolin- a. Dróżka pełznie mie˛dzy drzewami. (FM Tr)
guals (n ¼ 20; 13 female, 7 male; mean age ¼ 20.7; b. Rów pełznie mie˛dzy drzewami. (FM NTr)
age range 19–25); Polish L2 users of English c. Jaś pełznie mie˛dzy drzewami. (RM)
(n ¼ 20; 14 female, 6 male; mean age ¼ 22.1; age d. Dróżka jest mie˛dzy drzewami. (ST)

TABLE 1
Mean Self-Reported Proficiency

Speaking Understanding Spoken Language Reading

English monolinguals 9.9 9.85 9.75
L2 users in L2 English 8.2 8.3 9.0
L2 users in L1 Polish 9.7 9.8 9.6
Polish monolinguals 9.9 9.9 9.8
Note. 0 ¼ none; 10 ¼ highest
Ewa Tomczak and Anna Ewert 55

Each target sentence type made up 25% of all was created. The primary sources used for
critical targets in the study. To control for selecting 32 representative motion verbs in
potential lexical repetition priming effects, all English were Levin (1993) and Meteyard and
sentences of interest were pseudo-randomized Vigliocco (2009). Every target sentence of a
within the experimental blocks, with sentences particular sentence type was preceded by a
encoding the same verbs, paths, and grounds noun or a verb, with the experimental block
appearing every 32nd sentence. The stimulus containing each of the 64 sentences in one
presentation sequence was carefully computed, so priming condition only. Consequently, two
that every sentence type would appear as first, experimental trial blocks were designed, so that
second, third, and fourth a comparable number the distribution of prime (noun/verb)–target
of times. sentence pairs would be counterbalanced. Each
Prior to the experiment, a small-scale norming participant saw a given sentence in only one of
study was conducted that ensured that the critical the two priming conditions. Motion direction in
sentences were truly meaningful, semantically verb prime–sentence combinations was kept
plausible, and grammatical. A group of English constant, with the target sentences explicitly
monolinguals not participating in the experiment (RM) or implicitly (FM) conveying horizontal
was asked to complete a survey that featured 72 motion preceded by a horizontal verb prime,
sentences. The volunteers were to provide a rating and the target sentences with a vertical motion
on the meaningfulness of every single sentence on of the figure preceded by a vertical verb prime.
a fixed 7-point Likert scale, ranging from 1 (the For the Polish version of the experiment, transla-
sentence makes no sense to me at all) to 7 (the sentence tion equivalents of the primes in English were
makes perfect sense to me). To accept a given set of used.
sentences as target stimuli in the main experi-
Filler Sentences and Priming Stimuli. The
ment, a threshold of a minimum of 4 points was
remaining 64 sentences were adapted from
accepted a priori for each sentence in the set.
Curran et al. (1993), Dien et al. (2003), and
Only the lexical sets of sentences in which all four
Dien and O’Hare (2008) and served as semanti-
sentences obtained a minimum rating of 4 or
cally (e.g., The mole lived in a hole in the tax) and
above by every volunteer were selected as the
semantically–syntactically anomalous (e.g., The
primary target stimuli in the main study. Eventu-
earth is shaped like a develop) fillers. Sixty-four
ally, 16 sets were chosen as the critical target
pseudo-words, created in compliance with pho-
stimuli for the experiment. These were translated
notactic constraints in English for the English
into Polish by a native speaker of Polish, and then,
version and in Polish for the Polish version of the
with the help of the Poliqarp search engine,
experiment, served as prime stimuli for anoma-
checked against the data from The National Corpus
lous filler sentences.
of the Polish Language (430 million segments)
(Przepiórkowski et al., 2012). All translated pairs Design and Task Procedure. In each block and
of a noun (or a noun phrase) and a verb were each language, the participants went through the
checked for their possibility of co-occurrence in very same task procedures. Participants were
the Polish language. It was not possible to match randomly assigned to one of the two study blocks.
the noun–verb pairs on frequency and at the same The task across all versions of the experiment was
time maintain sentence equivalence in the two to decide whether sentences that were presented
language versions of the experiment, because the on the screen one by one made sense or did not
set of verbs that can be used in fictive motion make sense to them. They were asked to indicate
constructions is limited in both languages, as is their decisions on the computer keyboard by
the set of nontravelable entities that can enter pressing the “Z” key (does not make sense) or “/” key
into fictive motion constructions. (makes sense), with their left and right index
fingers, respectively. The experiment was pre-
Priming Stimuli. The study employed an un- ceded by a practice block.
masked priming paradigm, where each sentence Each experimental trial started with a screen-
(all sentences of interest and all fillers) was centred fixation cross (2000 ms), followed by a
preceded by a prime word that was congruent with prime (300ms), and then followed by a target
a target sentence with respect to language (i.e., sentence. Since the experiment employed a self-
English or Polish, respectively). A word list of 32 paced response paradigm, a target sentence
two- (or three-)syllable nouns, high in familiarity, remained on the screen until the participant
concreteness, and imageability (MRC Psycholin- performed the meaningfulness judgment. Partic-
guistic Database; Coltheart, 1981; Wilson, 1988) ipants’ judgments and RTs were recorded. The
56 The Modern Language Journal 99 (2015)

presentation of the 128th trial marked the end of sensible sentences (Marino et al., 2012). Figure 1
the experiment. shows the proportions (i.e., percentages) of
While both monolingual groups completed accepted sentences for all sentence types in
the sentence meaningfulness task in their native each of the language groups.
language, the Polish L2 users of English were To maintain data integrity, reaction time (RT)
asked to do the English version as well as responses above and below 3 standard deviations
the Polish version of the experiment on two from a participant’s mean (by condition) for
different occasions. The order of languages for L2 processing sentences accepted as meaningful
users (L1  L2, L2  L1) was counterbalanced. were excluded. This led to the exclusion of less
The pilot version of the experiment was carried than 1% of the data (outliers were evenly
out in each language with a group of five distributed across different groups and different
volunteers. conditions). Then, analyses of variance with
planned comparisons or Tukey’s HSD post hoc
tests (an alpha level set at .05) were carried out as
Data Analyses appropriate to test for (a) the presence of priming
We first analyzed the acceptability of the effects in the processing of four sentence types,
sentences. Analyses of variance with Tukey’s and (b) differences in the processing latencies
HSD post hoc tests (with an alpha level set at between different sentence types within and
.05) were run where applicable to explore between groups.
the differences in meaningfulness judgments
obtained for four sentence types within and RESULTS
between groups. As the data we obtained were
YES (1) and NO (0) decisions, all the analyses Sentence Meaningfulness Judgments
were carried out on the mean meaningfulness
ratings calculated per participant per sentence Figure 1 shows the mean sentence meaningful-
type. Because the assumption of homogeneity of ness ratings for the four sentence types across four
variance was met, no transformation was con- groups. A two-way (4  4) ANOVA (group  sen-
ducted on the data. The sentences rated as tence type) with a repeated measures design
meaningless were subsequently removed from produced a significant main effect of group, F(3,
the data set before response time analyses. The 228) ¼ 9.03, p < .0001, h2 ¼ .26, a significant main
rationale for this was that responses to sensible effect of sentence type, F(3, 228) ¼ 126.73,
action sentences are usually faster than to non- p < .0001, h2 ¼ .63, and a significant group 
sentence type interaction, F(9, 228) ¼ 11.97,
p < .0001, h2 ¼ .3.
Tukey’s HSD post hoc tests showed that mean
FIGURE 1
sentence meaningfulness ratings obtained for all
Mean Sentence Meaningfulness by Sentence Type
for all Groups
the sentences in the Polish monolingual group
(M ¼ .66) differed significantly from mean sen-
tence meaningfulness ratings obtained by the
English monolinguals (M ¼ .83, p < .001), Polish
L2 users of English in Polish (M ¼ .76, p < .05),
and Polish L2 users of English in English
(M ¼ .80, p < .001). The difference between
English monolinguals (M ¼ .83) and Polish L2
users of English in English (M ¼ .80) was not
significant (p > .05). The performance on mean-
ingfulness ratings of Polish L2 users of English in
English (M ¼ .80) did not differ significantly from
their performance in Polish (M ¼ .76, p > .05).
Tukey’s HSD post hoc tests for a significant
main effect of sentence type also revealed that
in all groups the FM NTr sentence type (M ¼
Note. Error bars represent standard errors; FM Tr ¼ .60) differed significantly from the FM Tr sen-
fictive motion with travelable entities; FM NTr ¼ fictive tence type (M ¼ .79, p < .0001), from RM senten-
motion with nontravelable entities; RM ¼ real motion; ces (M ¼ .76, p < .0001), and from ST sentences
ST ¼ static depictions. (M ¼ .89, p < .0001). ST sentences obtained
Ewa Tomczak and Anna Ewert 57

the highest sentence acceptability rates (M ¼ .89) ness ratings for the FM NTr sentence type (M ¼
and differed significantly from FM Tr sen- .75) differed significantly from the FM Tr
tences (M ¼ .79, p < .0001), FM NTr sentences sentence type (M ¼ .84, p < .05) and from ST
(M ¼ .60, p < .0001), and RM sentences (M ¼ .89, sentences (M ¼ .91, p < .001). Additionally, RM
p < .0001). Additionally, the difference between sentences (M ¼ .80) turned out to be less
mean meaningfulness ratings obtained for FM Tr meaningful for the participants than ST senten-
sentences and RM sentences was not significant ces, with the difference in the mean acceptability
(p > .05). rates reaching significance (M ¼ .91, p < .05).
Between-group analyses by sentence type (Tu- The Polish monolinguals showed a significant
key’s HSD post hoc tests) revealed that the two main effect of sentence type, F(3, 57) ¼ 93.80,
monolingual groups differed significantly in the p < .0001, h2 ¼ .83. Again, the FM NTr sentence
mean acceptability of FM NTr sentences, with type was observed to be the least accepted of all
English monolinguals (M ¼ .75) judging this type sentence types (M ¼ .35). The post hoc tests
as more acceptable than Polish monolinguals indicated that mean sentence acceptability rates
(M ¼ .35), p < .0001. All other between-group for the FM NTr sentence type (M ¼.35) differed
differences for particular sentence types (FM significantly from all the other sentence types: FM
Tr, RM, and ST sentences) between the two Tr (M ¼ .70, p < .001), RM (M ¼ .69, p < .001) and
monolingual groups were not significant. Tukey’s ST (M ¼ .88, p < .001). Additionally, mean accept-
HSD post hoc tests used to test for differences in ability of sentences was the highest for ST
mean sentence meaningfulness ratings by Polish sentences (M ¼ .88, p < .001), which differed
L2 users of English in each of their languages significantly from FM Tr (M ¼ .69, p < .001) and
showed that none of the between-group differ- from RM sentences (M ¼ .69, p < .001).
ences for sentences of the same type (FM Tr, FM Comparable analyses were conducted on data
NTr, RM, and ST sentences) were significant; obtained from the Polish L2 users of English in
however the L2 users accepted more FM senten- either of their languages. In English, a significant
ces (FM Tr and especially FM NTr) as meaningful main effect of sentence type, F(3, 57) ¼ 8.82,
in English than in Polish. Tukey’s HSD post hoc p < .0001, h2 ¼ .32 was found. FM NTr sentences
tests on mean sentence meaningfulness ratings obtained the lowest mean acceptability of all
between Polish L2 users in English and English investigated sentence types (M ¼ .71) and were
monolinguals revealed no significant differences found to differ significantly from two sentence
between the compared sentence types (FM Tr, FM types: FM Tr (M ¼ .83) at p < .01, and ST
NTr, RM, and ST). Finally, Tukey’s HSD post hoc (M ¼ .87) at p < .001, as revealed by Tukey’s
tests on mean sentence meaningfulness ratings HSD post hoc tests. The mean acceptability of
between Polish L2 users in Polish and Polish RM sentences (M ¼ .79) did not differ significant-
monolinguals showed that the mean acceptabi- ly from mean acceptability for FM NTr sentences,
lity of FM NTr sentences significantly differed nor from the mean acceptability of ST sentences.
between the groups, with Polish L2 users In the Polish language, a significant sentence type
(M ¼ .58) judging this type as more acceptable effect was obtained, F(3, 57) ¼ 42.05, p < .0001,
in Polish than Polish monolinguals (M ¼ .35), at h2 ¼ .69. Again, FM NTr sentences turned out to
p < .0001. No significant differences were found be the least accepted of all types (M ¼ .58). The
between Polish L2 users in Polish and Polish meaningfulness of the FM NTr type was signifi-
monolinguals for the remaining sentence types cantly lower than for FM Tr (M ¼ .80), RM
(FM Tr, RM, ST). (M ¼ .78), and ST (M ¼ .91), at p < .001 in all
To explore the effect of sentence type on cases. The highest mean acceptability of senten-
participants’ sentence meaningfulness judg- ces was reported for the ST type (M ¼ .91), which
ments in each group, a one-way repeated differed significantly from the FM Tr (M ¼ .69,
measures ANOVA was carried out. We provide p < .01) and RM type (M ¼ .69, p < .001).
the results of within-group analyses along with
mean meaningfulness ratings obtained in each Priming Effects
group for each particular sentence type.
In the group of English monolinguals, a Mean RTs in the different priming conditions
significant main effect of sentence type was for all the groups are presented in Figure 2. A
found, F(3, 57) ¼ 11.56, p < .0001, h2 ¼ .38, with three-way repeated measures ANOVA with a
FM NTr sentences being the least accepted of all 4  4  2 design was performed on mean RTs
sentence types (M ¼ .75). Tukey’s HSD post hoc across all the groups, to investigate the effects of
tests revealed that mean sentence meaningful- three independent variables (group, sentence
58 The Modern Language Journal 99 (2015)
FIGURE 2
Mean Reaction Times for Sentence Type by Priming Condition (Verb/Noun) Across Four Groups

3000

2500

2000
RT

1500

1000

500

0
FM Tr FM NTr RM ST FM Tr FM NTr RM ST FM Tr FM NTr RM ST FM Tr FM NTr RM ST
English monolinguals L2 users of English in English L2 users of English in Polish Polish monolinguals

Sentence Type
verb prime noun prime

Note. Error bars represent standard errors; FM Tr ¼ fictive motion with travelable entities; FM NTr ¼ fictive motion
with nontravelable entities; RM ¼ real motion; ST ¼ static depictions.

type, priming condition) on participants’ RTs. prime, noun prime) and four sentence types (FM
The main effect of group was not significant, F(3, Tr, FM NTr, RM, ST). The main effect of prime
76) ¼ 1.87, p ¼ .14, h2 ¼ .07. The main effect of was statistically significant, F(1, 19) ¼ 34.46,
sentence type was significant, F(3, 228) ¼ 61.5, p < .0001, h2 ¼ .64, as was the main effect of
p < .0001, h2 ¼ .45, and the main effect of prime sentence type, F(3, 57) ¼ 59.14, p < .0001,
was significant, F(3, 228) ¼ 23.47, p < .0001, h2 ¼ .76. Importantly, there was a significant
h2 ¼ .24. The group  prime interaction was not interaction between prime type and sentence
significant, F(3, 76) ¼ 2.11, p ¼ .11, h2 ¼ .08. The type, F(3, 57) ¼ 3.39, p < .05, h2 ¼ .15. Planned
sentence type  prime interaction was not signifi- pairwise comparisons revealed that mean pro-
cant, F(3, 228) ¼ .44, p ¼ .73, h2 ¼ .006. The cessing latency of FM Tr sentences was significan-
group  sentence type interaction was significant, tly shorter in the verb priming condition (M ¼
F(9, 228) ¼ 4.8, p < .0001, h2 ¼ .16. The group  2042.64 ms) than in the noun priming condition
sentence type  prime interaction was significant, (M ¼ 2278.00 ms), F(1, 19) ¼ 18.65, p < .001). FM
F(9, 228) ¼ 1.97, p < .05, h2 ¼ .07. The detailed NTr sentences were processed significantly faster
planned pairwise comparisons for a significant when preceded by a verb prime (M ¼ 2455.56 ms)
effect of prime showed that all sentence types than by a noun prime (M ¼ 2851.32 ms),
were processed faster when preceded by a verb F(1, 19) ¼ 12.33, p < .01. A statistically significant
(M ¼ 1980.97 ms) than by a noun (M ¼ 2108.18 difference was also found for RM sentences, F(1,
ms), F(1, 76) ¼ 23.47, p < .0001. The remaining 19) ¼ 12.85, p < .01, with the sentences being
relevant for our hypotheses planned pairwise processed faster in the verb priming condition
comparisons (or Tukey’s HSD post hoc tests (M ¼ 1709.05 ms) than in the noun priming
where more appropriate) for significant effects condition (M ¼ 1914.84 ms). The processing
and interactions for a three-way (4  4  2) time measures of ST depictions, as predicted,
repeated measures ANOVA are presented in the were not found to differ significantly between the
section Results for Processing by Sentence Type. conditions, F(1,19) ¼ 2.58, p ¼ .125, with M ¼
In the following section, we provide the results of 1748.30 ms for the verb priming condition and
within-group and between-group analyses. M ¼ 1825.01 ms when preceded by a noun prime.
English Monolinguals. The mean RT data were Polish Monolinguals. A two-way (2  4) repeated
subjected to a two-way (2  4) repeated measures measures ANOVA revealed a significant effect of
ANOVA, with two priming conditions (i.e., verb sentence type, F(3, 57) ¼ 6.75, p < .001, h2 ¼ .26.
Ewa Tomczak and Anna Ewert 59

No significant prime effect was found, F(1, FIGURE 3

19) ¼ .92, p ¼ .35, h2 ¼ .05, nor a prime  sentence Mean Reaction Times for Sentence Type Across Four
type interaction, F(3, 57) ¼ 1.70, p ¼ .18, h2 ¼ .08. Groups
We report the mean RTs (ms) in this group
per sentence type per priming condition: FM Tr
sentences (Mverb ¼ 2129.24, Mnoun ¼ 2070.76), FM
NTr sentences (Mverb ¼ 2240.38, Mnoun ¼ 2179.32),
RM sentences (Mverb ¼ 1868.18, Mnoun ¼ 1951.23),
ST sentences (Mverb ¼ 1871.97, Mnoun ¼ 2138.24).
Polish L2 Users of English in English. A two-way
(2  4) repeated measures ANOVA run on the
mean RT data revealed a significant effect
of sentence type, F(3, 57) ¼ 14.35, p < .0001,
h2 ¼ .43. The effect of prime was not significant,
with F(1, 19) ¼ 1.52, p ¼ .23, h2 ¼ .07, nor was the
prime  sentence interaction, F(3, 57) ¼ 1.29,
p ¼ .29, h2 ¼ .06. We report the mean RTs (ms)
in this group per sentence type per priming Note: Error bars represent standard errors; FM Tr ¼
condition: FM Tr sentences (Mverb ¼ 2161.81, fictive motion with travelable entities; FM NTr ¼ fictive
Mnoun ¼ 2373.42), FM NTr sentences (Mverb ¼ motion with nontravelable entities; RM ¼ real motion;
2435.49, Mnoun ¼ 2414.10), RM sentences (Mverb ¼ ST ¼ static depictions.
1908.59, Mnoun ¼ 1979.62), and ST sentences
(Mverb ¼ 1985.24, Mnoun ¼ 2048.24). was not significant, F(3, 76) ¼ 1.87, p ¼ .14,
Polish L2 Users of English in Polish. A two-way h2 ¼ .07. The detailed planned pairwise compar-
(2  4) repeated measures ANOVA produced two isons showed that, on the whole, there was a
significant main effects: the main effect of prime, significant difference between the processing
with F(1, 19) ¼ 12.61, p < .01, h2 ¼ .40, and the time of fictive motion sentences (FM) and real
main effect of sentence type, F(3, 57) ¼ 7.00, motion sentences, F(1, 76) ¼ 103.49, p < 0.0001.
p < .001, h2 ¼ .26. The prime  sentence type As we predicted, it took longer for the participants
interaction was not significant, F(3, 57) ¼ .92, to process both types of fictive motion sentences
p >.05, h2 ¼ .05. FM Tr sentences were processed (FM with M ¼ 2222.18 ms) than real motion
significantly faster in the verb condition (M ¼ sentences (RM with M ¼ 1830.95 ms). The com-
1700.16 ms) than in the noun condition (M ¼ puted planned pairwise comparisons also The
2002.20 ms), F(1, 19) ¼ 11.28, p < .01. A similar detailed planned pairwise comparisons for a
trend was observed for RM sentences, where the significant effect of prirevealed that FM Tr
difference was approaching significance (Mverb ¼ sentences (M ¼ 2094.78 ms) took longer to pro-
1595.05 ms, Mnoun ¼ 1721.07 ms, F(1, 19) ¼ 4.27, cess than RM sentences (M ¼ 1830.95 ms), F(1,
p ¼ .053). Neither for FM NTr nor for ST 76) ¼ 61.26, p < 0.0001. Analogically, FM NTr
sentences did the processing latencies differ sentences (M ¼ 2349.59 ms) took the participants
significantly in the two priming conditions more time to process than RM sentences (M ¼
(FM NTr: Mverb ¼ 2079.20 ms, Mnoun ¼ 2141.34 ms, 1830.95 ms), F(1, 76) ¼ 102.87, p < 0.0001.
F(1, 19) ¼ .17, p ¼ .69; ST: Mverb ¼ 1764.57 ms, We first present the results from separate one-
Mnoun ¼ 1842.09 ms, F(1, 19) ¼ .55, p ¼ .47). way repeated measures ANOVAs conducted in
each group, and then two-way (2  4) repeated
measures ANOVAs carried out to compare the
Results for Processing by Sentence Type means for processing sentence types between the
groups.
The mean RTs for the different sentence types
in all groups are presented in Figure 3. A three- English Monolinguals. A one-way repeated
way repeated measures ANOVA with a 4  4  2 measures ANOVA revealed a significant main
design performed on mean RTs across all the effect of sentence type, F(3, 57) ¼ 59.15, p < .0001,
groups produced a significant main effect of h2 ¼ .76. Planned pairwise comparisons revealed
sentence type, F(3, 228) ¼ 61.50, p < .0001, that FM sentences (including both FM Tr and FM
h2 ¼ .45, and a significant interaction between NTr sentences) were processed more slowly
sentence type and group, F(9, 228) ¼ 4.8, (M ¼ 2406.88 ms) than RM sentences (M ¼
p < .0001, h2 ¼ .16. The main effect of group 1811.94 ms), F(1, 19) ¼ 79.38, p < .0001, and
60 The Modern Language Journal 99 (2015)

more slowly than ST sentences (M ¼ 1786.66 ms), found for the processing of ST sentences was
F(1, 19) ¼ 68.66, p < .0001. not significant (M ¼ 1786.66 ms for English
monolinguals, M ¼ 2005.10 ms for Polish mono-
Polish Monolinguals. A one-way repeated meas-
linguals, p ¼ .64).
ures ANOVA showed a significant main effect of
sentence type, F(3, 57) ¼ 6.75, p < .001, h2 ¼ .26. Polish L2 Users of English in English vs. English
Planned pairwise comparisons revealed that FM Monolinguals. A two-way (2  4) repeated meas-
sentences (including both FM Tr and FM NTr ures ANOVA produced a significant main effect
sentences) were processed more slowly (M ¼ of sentence type, F(3, 114) ¼ 62.12, p < .0001,
2154.93 ms) than RM sentences (M ¼ 1909.71 h2 ¼ .62; and a significant sentence type  group
ms), F(1, 19) ¼ 14.28, p < .01, and more slowly interaction, F(3, 114) ¼ 6.41, p < .001, h2 ¼ .14.
than ST sentences (M ¼ 2005.10 ms), F(1, The effect of group was not significant, F(1,
19) ¼ 6.38, p < .05. 38) ¼ .15, p ¼.70, h2 ¼ .004. Tukey’s HSD post hoc
tests found no statistically significant differences
Polish L2 Users of English in English. A one-way
(at p < .05) in the mean processing times for each
repeated measures ANOVA produced a signifi-
sentence type (FM Tr, FM NTr, RM, ST) between
cant main effect of sentence type, F(3,
the groups.
57) ¼ 14.35, p < .0001, h2 ¼ .43. Planned pairwise
comparisons showed that FM sentences (includ- Polish L2 Users of English in Polish vs. Polish
ing both FM Tr and FM NTr sentences) were Monolinguals. A two-way (2  4) repeated meas-
processed more slowly (M ¼ 2346.20 ms) than RM ures ANOVA revealed a significant main effect of
sentences (M ¼ 1944.11 ms), F(1, 19) ¼ 17.19, sentence type, F(3, 114) ¼ 12.77, p < .0001,
p < .001, and more slowly than ST sentences h2 ¼ .25. The effect of group was not significant,
(M ¼ 2016.74 ms), F(1, 19) ¼ 21.33, p < .001. F(1, 38) ¼ 2.82, p ¼ .10, h2 ¼ .07. The sentence
type  group interaction was not significant, F(3,
Polish L2 Users of English in Polish. A one-way
114) ¼ .65, p ¼.58, h2 ¼ .02.
repeated measures ANOVA showed a significant
main effect of sentence type, F(3, 57) ¼ 6.70, Polish L2 Users of English in English vs. Polish L2
p < .001, h2 ¼ .26. Planned pairwise comparisons Users of English in Polish. A two-way (2  4)
revealed that FM sentences (including both FM repeated measures ANOVA showed a significant
Tr and FM NTr sentences) were processed more main effect of sentence type, F(3, 114) ¼ 18.62,
slowly (M ¼ 1980.72 ms) than RM sentences (M ¼ p < .0001, h2 ¼ .33. The effect of group was not
1658.06 ms), F(1, 19) ¼ 18.64, p < .001, and more significant, F(1, 38) ¼ 3.56, p ¼.07, h2 ¼ .09. The
slowly than ST sentences (M ¼ 1803.33 ms), but in sentence type  group interaction was not signifi-
the latter case the difference was not significant, cant, F(3, 114) ¼ .81, p ¼.49, h2 ¼ .02.
F(1, 19) ¼ 2.86, p ¼ .11.
DISCUSSION
English Monolinguals vs. Polish Monolinguals.
A two-way (2  4) repeated measures ANOVA At the outset, we examined the meaningfulness
carried out to compare mean RTs for sentence judgments made by the L2 users and the
types between the two monolingual groups monolingual participants. Since the English and
produced a significant main effect of sentence Polish sentences had equivalent wording, it is
type, F(3, 114) ¼ 53.17, p < .0001, h2 ¼ .58; and a possible to make cross-linguistic comparisons to
significant sentence type  group interaction, see how the L2 users represent motion in both L2
F(3, 114) ¼ 15.90, p < .0001, h2 ¼ .29. The effect and L1. We established that the two monolingual
of group was not significant, F(1, 38) ¼ .19, groups differ in how they assess the meaningful-
p ¼ .66, h2 ¼ .01. Tukey’s HSD post hoc tests ness of the motion sentences used in this
revealed that English monolinguals processed experiment, with the monolingual Polish partic-
FM NTr sentences more slowly (M ¼ 2653.44 ms) ipants rating all the sentences lower on meaning-
than Polish monolinguals (M ¼ 2209.85 ms), at fulness. The largest difference in meaningfulness
p < .05. The difference found for the processing ratings between the monolingual groups con-
of FM Tr sentences was not significant (M ¼ cerns sentences with fictive motion with non-
2160.32 ms for English monolinguals, M ¼ travelable entities (i.e., FM NTr), but both types of
2100.00 ms for Polish monolinguals, p ¼ .99). fictive motion received lower ratings in the Polish
The difference observed for the processing of monolingual group. Although the monolingual
RM sentences was not significant (M ¼ 1811.94 ms groups differed in their judgments of sentence
for English monolinguals, M ¼ 1909.71 ms for meaningfulness, the L2 users rated the sentences
Polish monolinguals, p ¼ .99). The difference similarly in both languages. They rated the
Ewa Tomczak and Anna Ewert 61

sentences generally higher in English than in occur in the Polish monolingual speakers, and
Polish, but this difference was not statistically second, what exactly is being transferred from the
significant. Their L2 ratings were similar to the L2 to the L1 by the L2 users? One possibility is that
ratings of monolingual speakers of English, but the Polish monolingual group did not display the
rating the same sentences in the L1 they assessed expected behavior because of lower acceptability
them as more meaningful than the monolingual of these sentences. This, however, does not
speakers of Polish did. We interpret this as an explain why the L2 users do not display this
instance of transfer from L2 to L1. behavior in their L2 English either, even though
The FM NTr sentences were found to be the they rate the sentences as high as monolingual
least acceptable to all the groups of participants, speakers of English do and display it in their L1
while sentences with fictive motion with travelable Polish.
entities (i.e., FM Tr) were rated similarly to RM Another possibility is that speakers of Polish
sentences in all the groups. The FM NTr have different sentence processing strategies
sentences were the least acceptable to monolin- from speakers of English, Polish being a free
gual speakers of Polish. This is surprising, word order language. Previous research (Mishra
considering that Polish as a language with higher & Singh, 2010) suggests that speakers of free word
manner salience than English (cf. Czechowska & order languages might pay more attention to the
Ewert, 2011) should have fewer restrictions on the figure than to the verb while processing motion
participation of different figures in fictive motion descriptions, hence they will not pay enough
events (Matsumoto, 1996). This result might have attention to the verb for the verb primes to have
been induced by the choice of the experimental any effect. They might focus more on the figure
sentences. and its capacity for action of a particular kind than
All the predictions stemming from the ground- on the features of motion being described. If so,
ing hypothesis were confirmed in the group of perhaps speakers of a free word order language
monolingual speakers of English, who demon- simply consider the sentence constituent that
strated the predicted priming effect in the comes first in the sentence to be the most
processing of all motion sentences. This is important for interpreting the meaning of the
consistent with previous findings on English L1 sentence, or else, perhaps they have richer
speakers (e.g., Glenberg & Kaschak, 2002; Zwaan conceptual representations of the capacity of
& Taylor, 2006). However, none of the predic- figures for engaging in motion of different kinds.
tions stemming from this hypothesis could be Yet another possibility is that Polish speakers
confirmed with monolingual speakers of Polish. ground motion differently in the brain, their
The most interesting results in this regard were mental representations of motion as encoded in
obtained with the L2 users. The L2 users the verb being more abstract. This possibility is
demonstrated no priming effect in their L2 suggested by the morphological complexity of the
English. In line with previous research this should Polish verb. Research on morphological process-
be interpreted as transfer from the L1 to the L2. ing indicates that the semantic meaning of
However, they did demonstrate some instances of derivational morphemes is accessed very early in
transfer from the L2 to their L1, generally word recognition (Feldman et al., 2009, 2012).
responding to verb primes in their processing Many Polish verbal prefixes are ambiguous in
of motion sentences in Polish. The priming meaning and this ambiguity has to be resolved at a
effect was significant for FM Tr sentences, and very early stage of word processing (cf. Tsang &
approached significance for RM sentences. The Chen, 2013, for explication of how ambiguous
effect was not significant for FM NTr sentences, morphemes may affect priming effects). Also De
which might be related to generally lower Grauwe et al. (2014) in a recent fMRI study
acceptability of these sentences. In our discussion demonstrate lack of embodiment effects for
of previous research on embodied cognition, the opaque complex verbs in both native and L2
possibility was raised that L2 users may not activate speakers.
their action representations in the L2. This Whichever is the case, the L2 users’ responsive-
question now becomes immaterial since Polish ness to verb primes in their L1 processing of
monolinguals do not seem to activate their action motion sentences constitutes a clear instance of
representations either. What is interesting is that transfer of behavioral patterns. However, they do
the L2 users do activate those representations in not respond to verb primes while processing
their L1. motion sentences in their L2, which indicates
Two questions require further exploration: transfer of processing preferences from the L1 to
First, why did the expected priming effect not the L2. This bidirectional transfer in L2 users’
62 The Modern Language Journal 99 (2015)

response latencies suggests that explanations languages might be modulated by the language
based on either morphological or word order they speak, L2 users will display a highly complex
processing, available in the embodiment litera- pattern of responses that shows influences from
ture so far, are not sufficient to explain the either of their languages. While the differences
differences in how English and Polish mono- between the monolingual groups in our study can
linguals ground motion. It is probably a combi- be accounted for on the basis of previous research
nation of these two factors that accounts for the in the field, we demonstrate that research with L2
differences. users can unravel even further intricacies.
The predictions of the simulation hypothesis The study’s L2 users were L1 dominant and
were fully confirmed: Participants in all groups resident in an L1 dominated environment at the
processed both types of fictive motion longer than time of the experiment. Even so, the L2 turned
static depictions and real motion. Between-group out to have a pervasive influence on their
comparisons yielded only one significant differ- cognitive functioning, affecting both L1 sentence
ence for processing particular sentence types: meaningfulness ratings and reactions to verb
English monolinguals processed FM NTr senten- primes in a psycholinguistic experiment. This
ces more slowly than Polish monolinguals. This confirms that L2 users have different minds (cf.
difference most likely results from different Cook, 1992) and cannot be expected to behave
acceptability rates for these sentences in the like monolinguals in all respects.
respective groups. The Polish monolinguals The L2 users were highly proficient students of
rejected more FM NTr sentences, but they English who participate in an all-English program
responded faster to those they considered seman- at the university where they receive instruction
tically meaningful. The English monolinguals and take course exams in English. Their high
accepted more sentences as meaningful, but took proficiency in English, reflected in their self-
longer to respond to the ones they considered ratings, is confirmed by their experimental
acceptable. results: They judged the acceptability of the
experimental sentences similarly to the American
CONCLUSIONS AND IMPLICATIONS FOR L2 students and their response times in the whole
LEARNING experiment were also similar. Despite this high
L2 proficiency, they evidently processed the
This study examined how L2 users represent English sentences differently from the American
and process real and fictive motion in L1 and L2. students, as demonstrated by a lack of the
The study’s L2 users demonstrated reverse expected priming effects in their responses to
transfer, with their L2 influencing their sentence the English sentences. One might interpret this
meaningfulness ratings in the L1. We also tested to mean that L1 processing preferences are very
whether the findings from motion and action difficult to change even in highly advanced L2
research conducted with monolinguals in the learners. However, that interpretation would
embodied cognition perspective applied as well to disregard the fact that the study participants at
monolinguals speaking different languages and the same time showed English-like processing
to L2 users. Our findings show that the partic- styles in their L1 Polish. Another possibility is
ipants processed figurative motion sentences suggested by Barsalou et al. (2008), namely that
similarly, but there were differences in responses in the L2 they engage in shallow, linguistic
to verb primes between monolingual speakers of processing only, not accessing the motor repre-
different languages that implicate differences in sentations of motion. If this is the case, it means
conceptual processing. With regard to the effect that their comprehension of spatial relations and
of verb primes on sentence processing, the study’s movement in space is impoverished. In other
L2 users demonstrated bidirectional transfer, words, they understand the language, but it does
behaving in the L1 like the experiment’s English not evoke as rich spatial and temporal imagery as
monolinguals and demonstrating behaviors simi- it would in native speakers, which might be vital
lar to the study’s Polish monolinguals in their L2. to understanding some specific problems de-
As the conceptual priming experiment scribed in the L2. A similar problem is experi-
explored the relation of linguistic processing to enced by children who are learning to read in
representations of physical action, our findings their native language. Glenberg et al. (2007)
show once again that human cognition cannot demonstrate that activities involving the manipu-
be studied with monolingual speakers only (cf. lation of objects in a way that corresponds to the
Henrich, Heine, & Norenzayan, 2010). While the sentences they read or even imagining the
responses of monolingual speakers of different manipulation improve reading comprehension
Ewa Tomczak and Anna Ewert 63

in reading intervention programs. The results Chatterjee, A. (2010). Disembodying cognition. Lan-
of our priming experiment suggest that even guage and Cognition, 2, 79–116.
advanced L2 learners may benefit from such Coltheart, M. (1981). The MRC Psycholinguistic Data-
techniques while struggling with spatio–temporal base. Quarterly Journal of Experimental Psychology,
33A, 497–505.
problems worded in the L2.
Cook, V. (1992). Evidence for multicompetence.
Language Learning, 42, 557–591.
Curran, T., Tucker, D. M., Kutas, M., & Posner, M. I.
ACKNOWLEDGMENTS (1993). Topography of the N400: Brain electrical
activity reflecting semantic expectancy. Electroen-
cephalography and Clinical Neurophysiology, 88, 188–
The research reported in this article was partly 209.
supported by grant number N N104 380040 from the Czechowska, N., & Ewert, A. (2011). Perception of
Polish Ministry of Science and Higher Education and motion by Polish–English bilinguals. In V. Cook &
the National Science Centre to Anna Ewert. We are B. Bassetti (Eds.), Language and bilingual cognition
grateful to the editors and two anonymous reviewers for (pp. 287–314). New York/Hove, UK: Psychology
insightful comments on earlier versions of this article. Press.
We are also grateful to Libby Damjanovic for insightful De Grauwe, S., Willems, R. M., Rueschemeyer, S.–A.,
comments on the analyses reported here during the Lemhöfer, K., & Schriefers, H. (2014). Embodied
AAAL 2012 conference in Boston. language in first- and second-language speakers:
Neural correlates of processing motor verbs.
Neuropsychologia, 56, 334–349.
Desai, R. H., Conant, L. L., Binder, J. R., Park, H.,
REFERENCES & Seidenberg, M. S. (2013). A piece of the
action: Modulation of sensory–motor regions by
Athanasopoulos, P., & Bylund, E. (2013). Does gram- action idioms and metaphors. Neuroimage, 83, 862–
matical aspect affect motion event cognition? A 869.
cross-linguistic comparison of English and Swed- Dien, J., Frishkoff, G. A., Cerbone, A., & Tucker,
ish speakers. Cognitive Science, 37, 286–309. D. M. (2003). Parametric analysis of event-related
Barsalou, L. W. (1999). Perceptual symbol systems. potentials in semantic comprehension: Evidence
Behavioral and Brain Sciences, 22, 577–660. for parallel brain mechanisms. Cognitive Brain
Barsalou, L. W. (2008). Grounded cognition. Annual Research, 15, 137–153.
Review of Psychology, 59, 617–645. Dien, J., & O’Hare, A. J. (2008). Evidence for automatic
Barsalou, L. W., Santos, A., Simmons, W. K., & Wilson, sentence priming in the fusiform semantic
C. D. (2008). Language and simulation in area: Convergent ERP and fMRI findings. Brain
conceptual processing. In M. De Vega, A. M. Research, 1243, 134–145.
Glenberg, & A. C. Graesser (Eds.), Symbols, Feldman, L. B., Kostić, A., Gvozdenović, V., O’Connor,
embodiment, and meaning (pp. 245–283). Oxford: P. A., Moscoso del Prado, J., & Martı́n, F. (2012).
Oxford University Press. Semantic similarity influences early morphologi-
Bedny, M., & Caramazza, A. (2011). Perception, action, cal priming in Serbian: A challenge to form-then-
and word meanings in the human brain: The case meaning accounts of word recognition. Psycho-
from action verbs. Annals of the New York Academy of nomic Bulletin & Review, 19, 668–676.
Sciences, 1224, 81–95. Feldman, L. B., O’Connor, P. A., Moscoso del Prado, J.,
Bergen, B., Lau, T.–T. C., Narayan, S., Stojanovic, D., & & Martı́n, F. (2009). Early morphological process-
Wheeler, K. (2010). Body part representations in ing is morphosemantic and not simply morpho–
verbal semantics. Memory & Cognition, 38, 969–981. orthographic: A violation of form-then-meaning
Brown, A., & Gullberg, M. (2008). Bidirectional accounts of word recognition. Psychonomic Bulletin
crosslinguistic influence in L1–L2 encoding of & Review, 16, 684–691.
manner in speech and gesture: A study of Japanese Flecken, M. (2011). Event conceptualization by early
speakers of English. Studies in Second Language Dutch–German bilinguals: Insights from linguistic
Acquisition, 30, 225–251. and eye-tracking data. Bilingualism: Language and
Brown, A., & Gullberg, M. (2010). Changes in encoding Cognition, 14, 61–77.
of PATH of motion in a first language during Glenberg, A. M., Jaworski, B., Rischal, M., & Levin, J. R.
acquisition of a second language. Cognitive Lin- (2007). What brains are for: Action, meaning, and
guistics, 21, 263–286. reading comprehension. In D. McNamara (Ed.),
Bylund, E. (2009). Effects of age of L2 acquisition on L1 Reading comprehension strategies: Theories, interven-
event conceptualization patterns. Bilingualism: tions, and technologies (pp. 221–240). Mahwah, NJ:
Language and Cognition, 12, 305–322. Lawrence Erlbaum.
Bylund, E., & Jarvis, S. (2011). L2 effects on L1 event Glenberg, A. M., & Kaschak, M. P. (2002). Grounding
conceptualization. Bilingualism: Language and language in action. Psychonomic Bulletin & Review,
Cognition, 14, 47–59. 9, 558–565.
64 The Modern Language Journal 99 (2015)
Henrich, J., Heine, S. J., & Norenzayan, A. (2010). The (pp. 1133–1138). Austin, TX: Cognitive Science
weirdest people in the world? Behavioral and Brain Society.
Sciences, 33, 61–83. Przepiórkowski, A., Bańko, M., Górski, R. L., &
Hohenstein, J., Eisenberg, A., & Naigles, L. (2006). Is he Lewandowska–Tomaszczyk, B. (2012). Narodowy
floating across or crossing afloat? Cross-influence Korpus Je˛zyka Polskiego, [The National Corpus of the
of L1 and L2 in Spanish–English bilingual adults. Polish Language]. Warsaw, Poland: Wydawnictwo
Bilingualism: Language and Cognition, 9, 249–261. Naukowe PWN.
Hostetter, A. B., & Alibali, M. W. (2008). Visible Pulvermüller, F., Härle, M., & Hummel, F. (2001).
embodiment: Gestures as simulated action. Psy- Walking or talking? Behavioral and neurophysio-
chonomic Bulletin & Review, 15, 495–514. logical correlates of action verb processing. Brain
Levelt, W. J. M. (1989). Speaking: From intention to and Language, 78, 143–168.
articulation. Cambridge, MA: The MIT Press. Richardson, D. C., & Matlock, T. (2007). The integra-
Levin, B. (1993). English verb classes and alternations: A tion of figurative language and static depictions:
preliminary investigation. Chicago: University of An eye movement study of fictive motion. Cogni-
Chicago Press. tion, 102, 129–138.
Marian, V., Blumenfeld, H. K., & Kaushanskaya, M. Rojo, A., & Valenzuela, J. (2009). Fictive motion in
(2007). The Language Experience and Proficiency Spanish: Travellable, non-travellable, and path-
Questionnaire (LEAP–Q): Assessing language related manner information. In J. Valenzuela, A.
profiles in bilinguals and multilinguals. Journal of Rojo, & C. Soriano (Eds.), Trends in cognitive
Speech, Language and Hearing Research, 50, 940–967. linguistics: Theoretical and applied models (pp. 244–
Marino, B. F., Gallese, V., Buccino, G., & Riggio, L. 260). Hamburg, Germany: Peter Lang.
(2012). Language sensorimotor specificity modu- Schmiedtová, B., von Stutterheim, C., & Carroll, M.
lates the motor system. Cortex, 48, 849–856. (2011). Implications of language-specific patterns
Matlock, T. (2004a). The conceptual motivation of in event construal of advanced L2 speakers. In A.
fictive motion. In G. Radden & K.–U. Panther Pavlenko (Ed.), Thinking and speaking in two
(Eds.), Studies in linguistic motivation (pp. 221– languages (pp. 66–107). Bristol, UK: Multilingual
248). Berlin: Mouton de Gruyter. Matters.
Matlock, T. (2004b). Fictive motion as cognitive Simmons, W. K., Hamann, S. B., Harenski, C. N., Hu,
simulation. Memory & Cognition, 32, 1389–1400. X. P., & Barsalou, L. W. (2008). FMRI evidence for
Matlock, T. (2006). Depicting fictive motion in draw- word association and situated simulation in
ings. In J. Luchjenbroers (Ed.), Cognitive linguis- conceptual processing. Journal of Physiology–Paris,
tics: Investigations across languages, fields, and 102, 106–119.
philosophical boundaries (pp. 67–85). Philadel- Slobin, D. I. (1996). From “thought and language” to
phia/Amsterdam: John Benjamins. “thinking for speaking.” In J. J. Gumperz & S. C.
Matlock, T., & Richardson, D. C. (2004). Do eye Levinson (Eds.), Rethinking linguistic relativity (pp.
movements go with fictive motion? In K. D. 70–96). Cambridge: Cambridge University Press.
Forbus, D. Gentner, & T. Reiger (Eds.), Proceedings Slobin, D. I. (2006). What makes manner of motion
of the 26th Annual Conference of the Cognitive Science salient? Explorations in linguistic typology, dis-
Society (pp. 909–914). Mahwah, NJ: Lawrence course, and cognition. In M. Hickmann & S.
Erlbaum. Robert (Eds.), Space in languages: Linguistic systems
Matsumoto, Y. (1996). Subjective motion and English and cognitive categories (pp. 59–81). Philadelphia/
and Japanese verbs. Cognitive Linguistics, 7, 183– Amsterdam: John Benjamins.
226. Talmy, L. (1985). Lexicalization patterns: Semantic
Meteyard, L., & Vigliocco, G. (2009). Verbs in space: Axis structure in lexical forms. In T. Shopen (Ed.),
and direction of motion norms for 299 English Language typology and syntactic description (Vol. 3,
verbs. Behavior Research Methods, 41, 565–574. pp. 57–149). Cambridge: Cambridge University
Mishra, R. K., & Singh, N. (2010). Online fictive motion Press.
understanding: An eye-movement study with Talmy, L. (2000a). Toward a cognitive semantics, Vol. 1
Hindi. Metaphor and Symbol, 25, 144–161. Cambridge, MA: The MIT Press.
Pavlenko, A. (2010). Verbs of motion in L1 Russian of Talmy, L. (2000b). Toward a cognitive semantics, Vol. 2
Russian–English bilinguals. Bilingualism: Language Cambridge, MA: The MIT Press.
and Cognition, 13, 49–62. Talmy, L. (2013). Concept structuring systems in
Pavlenko, A., & Volynsky, M. (2015). Motion encoding in language. In M. Tomasello (Ed.), The new
Russian and English: Moving beyond Talmy’s psychology of language: Cognitive and functional
typology. Modern Language Journal, 99, Supplement approaches to language structure (pp. 15–46). Mah-
2015, 32–48. wah, NJ: Lawrence Erlbaum.
Perniss, P., Vinson, D., Fox, N., & Vigliocco, G. (2013). Tettamanti, M., Buccino, G., Saccuman, M. C., Gallese,
Comprehending with the body: Action compati- V., Danna, M., Scifo, P., et al. (2005). Listening to
bility in sign language? In M. Knauff, M. Pauen, N. action-related sentences activates fronto–parietal
Sebanz, & I. Wachsmuth (Eds.), Proceedings of the motor circuits. Journal of Cognitive Neuroscience, 17,
35th Annual Conference of the Cognitive Science Society 273–281.
Ewa Tomczak and Anna Ewert 65
Thierry, G., Athanasopoulos, P., Wiggett, A., Dering, B., Doehler (Eds.), EuroSLA Yearbook (pp. 183–206).
& Kuipers, J.–R. (2009). Unconscious effects of Philadelphia/Amsterdam: John Benjamins.
language-specific terminology on preattentive Whorf, B. L (1940/1956). Linguistics as an exact science.
color perception. Proceedings of the National Acade- In J. B. Carroll (Ed.), Language, thought, and reality.
my of Sciences, 106, 4567–4570. Selected writings of Benjamin Lee Whorf (pp. 220–232).
Tsang, Y.–K., & Chen, H.–C. (2013). Early morphologi- Cambridge, MA: The MIT Press.
cal processing is sensitive to morphemic mean- Wilson, M. (2002). Six views of embodied cognition.
ings: Evidence from processing ambiguous Psychonomic Bulletin & Review, 9, 625–636.
morphemes. Journal of Memory and Language, 68, Wilson, M. (1988). The MRC Psycholinguistic Database:
223–239. Machine Readable Dictionary, Version 2. Behav-
Tucker, M., & Ellis, R. (2004). Action priming by briefly iour Research Methods, Instruments, & Computers, 20,
presented objects. Acta Psychologica, 116, 185– 6–11.
203. Zwaan, R. A., & Taylor, L. J. (2006). Seeing, acting,
von Stutterheim, C. (2003). Linguistic structure and understanding: Motor resonance in language
information organisation. The case of very ad- comprehension. Journal of Experimental Psychology:
vanced learners. In S. Foster–Cohen & S. Pekarek– General, 135, 1–11.