Grammatical Aspect influences Event Duration Estimations: Evidence from Dutch

Abstract

This study investigates the effect of grammatical aspect marking in Dutch sentences, on speakers' estimations of the duration of highly familiar, everyday events. We first established the 'inherent' or natural duration of different events (Exp. 1). This was then used for the manipulation of aspect (Exp. 2). Participants dragged a slider across the computer screen to estimate the duration of progressive and non-progressive event descriptions. Findings show how the progressive form extends duration estimations for short events, whereas it shortens the perceived duration of inherently medium and long events. We interpret this as psycholinguistic evidence for the function of aspect in Dutch, i.e., giving an 'inside' view of the event and focusing a specific internal time span of the event.

Full text

Grammatical Aspect influences Event Duration Estimations:
Evidence from Dutch
Monique Flecken (m.flecken@donders.ru.nl)
Donders Centre for Cognition, Radboud University,
PO Box 9104, 6500 HE Nijmegen, the Netherlands
Johannes Gerwien (jo.gerwien@uni-heidelberg.de)
Institut für Deutsch als Fremdsprachenphilologie, Heidelberg University
Plöck 55, 69117 Heidelberg, Germany
Abstract
This study investigates the effect of grammatical aspect
marking in Dutch sentences, on speakers’ estimations of the
duration of highly familiar, everyday events. We first
established the ‘inherent’ or natural duration of different
events (Exp. 1). This was then used for the manipulation of
aspect (Exp. 2). Participants dragged a slider across the
computer screen to estimate the duration of progressive and
non-progressive event descriptions. Findings show how the
progressive form extends duration estimations for short
events, whereas it shortens the perceived duration of
inherently medium and long events. We interpret this as
psycholinguistic evidence for the function of aspect in Dutch,
i.e., giving an ‘inside’ view of the event and focusing a
specific internal time span of the event.
Keywords: grammatical aspect, event representation, Dutch.
Introduction
Time is an important domain of human experience. For
example, most people are able to roughly estimate how long
it takes to open a window, to prepare a certain meal or to
watch a movie, given normal circumstances. This
information about the time course of events is part of world
knowledge and our experience with different events and
situations. When people talk about events (in finite
sentences) the grammar of the language they speak may
require them to make specific distinctions which relate to
time explicit. They may be required to provide information
on whether an event is taking place in the present, or took
place in the past (grammatical category of tense). People
may also need to specify whether an event has just begun, is
in progression, or has reached a state of completion
(grammatical category of aspect). However, it is not clear,
in what ways world knowledge about temporal features of
events and the distinctions provided by the language system
interact: how do specific linguistic structures influence the
way people represent events? In this study we address this
question, and ask specifically how the use of aspectual verb
forms in a sentence context affects people’s general
knowledge about the temporal contours of events, i.e., the
duration of events.
Background
Linguistic theories on grammatical aspect (also viewpoint or
verbal aspect) state that the function of progressive aspect is
to modulate the inherent temporal contours of an event,
thereby defocusing its boundaries (e.g., Comrie, 1976; Dahl,
2000). Specifically, it expresses a particular perspective on
an event in that it is represented as a specific ‘ongoing’
instance of an event: For example, the semantic difference
between ‘he passes the ball’ and ‘he is passing the ball’. The
progressive defocuses the boundaries of the event, to give
an ‘inside’ view of a situation and thus ‘highlight’ its
intermediate phases (e.g., Comrie, 1976). It is important to
note that event descriptions that mark information regarding
tense or aspect (‘finite’ expressions) do not directly refer to
the time span defined by inherent temporal features of an
event. With regard to aspect, Klein (1994) for example,
distinguishes two temporal layers in language and describes
aspect as denoting the relation between the linguistically
unspecified time of an event (Time of Situation, TSit), and
the specific time span that is being talked about (Topic
Time, TT). The function of progressive aspect is to express
that this time span (TT) falls within the boundaries of the
event (Figure 1). This means that the time span at issue will
be viewed as having extended duration (event marked as in
progress), but it will be shorter than that of the entire event,
as the time span in question does not include the boundary
phases. Events not marked for progressive aspect, on the
other hand, are unspecific in this regard and can include the
entire event (‘passes the ball’), thus highlighting a
qualitatively different time span compared to events marked
for progressive aspect.
TT:
phase in progress
TSit: ‘to pass a ball’
‘He is passing the ball’
Figure 1: Time-relational analysis of progressive aspect
(cf. Klein, 1994)
The present study addresses the psycholinguistic reality of
the above claims on the function of aspect in a sentence
context: how exactly does this grammatical structure
influence the way in which events, as expressed by verbal

predicates, are perceived? We focus on potential
modulations of the perceived inherent or natural duration of
events.
Initial steps in understanding the role of grammatical
aspect in event conceptualization have been made from a
psycholinguistic perspective in production and
comprehension studies. A production study, comparing
mono-/bilingual speakers of aspect and non-aspect
languages, has looked at event descriptions and patterns in
gaze allocation (measured with eye tracking), while subjects
were preparing to speak about causative events (which
involve an agent acting on an object, e.g., a person knitting a
scarf) (Flecken, 2011a). Speakers who used progressive
aspect to describe the events (English, Dutch),
predominantly allocated gaze to features of the ongoing
action (the knitting), and less to the agent of the event
(which was the German pattern, non-aspect language).
Progressive aspect thus focuses visual attention to ongoing
aspects of an event, online, in production.
Comprehension studies have focused on the role of
grammatical aspect for our understanding of situations or
events, and the relations between different events. Magliano
and Schleich (2000), for example, show how grammatical
aspect constrains mental models of situations, when
connected within a narrative structure. In their comparison
of readers’ comprehension of sentences marked for
progressive and perfective aspect in the past tense, and
embedded in a stretch of discourse, they found differences
between comprehenders’ conceptions of events, despite use
of the same lexical information. When reading ‘Betty was
delivering a baby’ versus ‘Betty delivered a baby’ two
different mental representations of the event were formed
with consequences for the way in which further contextual
information was understood. Using a question-answer
paradigm, they explicitly asked whether the critical events
were finished or not, at specific points in the story line. In
one experiment, they addressed the question whether
‘general knowledge’ on the duration of events interacted
with aspect marking. They included events with a long and a
short duration (long duration: ranging from ‘watching a
movie’ to ‘writing a novel’; short duration: ranging from
‘scratching your nose’ to ‘packing a suitcase’). Likelihood
scores indicated that ‘long’ events, marked as in progress by
means of progressive aspect, were still perceived as ongoing
at later sentence positions, in contrast to ‘short’ events. As
they used a rather course measure (yes/no questions), we
cannot be sure how exact this difference for aspect marking
between short and long events is. Furthermore, the events
within each category showed a great range in duration
ratings, and included events that may not be familiar to all
participants (such as ‘giving birth’). A person’s lack of
experience with a situation or action may result in a less
precise mental model of the event. Their findings may be
interpreted as showing that the duration of the event
described with progressive aspect is interpreted as
prolonged, in comparison to the same event described by
non-progressive verb forms.
Madden & Zwaan (2003) also show how verbal aspect
constrains speakers’ representations of events. In a sentence
picture matching task, with pictures showing events at
different phases, they found that sentences marked with
progressive aspect (in the past tense) elicit an equal amount
of choices for pictures showing a completed or an
incompleted event. The authors interpret this as showing
that speakers can represent different phases of an event as in
progression.
Bergen & Wheeler (2010) also study the effect of aspect
on ‘mental simulation’. They find that speakers mentally
simulate the nucleus of an action, when described in English
sentences marked with progressive aspect, in contrast to
sentences with perfective aspect.
In, e.g., Anderson et al. (2008) a different methodology
was adopted, aiming to get a closer look at online
processing of aspectually marked event sentences. They
used a mouse tracking paradigm, in which speakers were
asked to place a figure on a path, on its way to an endpoint,
when listening to sentences describing motion events with
and without progressive aspect in the past tense (‘was
walking to school’ versus ‘walked to school’). Figures were
placed closer to the goal of the motion in the non-aspect
condition, indicating that the past progressive focuses
attention on internal phases of the past event.
These experiments provide important insights, as they
reveal more clearly how aspect influences the processing of
event structure. Important questions remain, however: For
example, how does progressive aspect modulate event
duration estimations for different event types?
In the present study, we take Dutch as our test case, as
there is the advantage that this language allows use of
sentences describing events in the present tense, both with
and without morphological marking of progressive aspect.
Production studies on Dutch have shown how progressive
aspect is used frequently, but not for all event types. Unlike
in English, use is not obligatory in any context (von
Stutterheim, Carroll & Klein, 2009; see for acceptability
judgements of progressive and non-progressive event
descriptions, Flecken, 2011b). With the investigation of a
language other than English, we set out to explore whether
the temporal relations described above for progressive
aspect (Comrie, 1976; Klein, 1994) apply when Dutch
speakers use the progressive aan het construction. In
linguistic terms, progressives in different languages will
follow the same temporal logic; but do speakers’ responses
reflect their role so as to modulate their perception of the
internal phases of a dynamic situation when estimating
event duration? Dutch is a language in which use of
progressive aspect is not fully grammaticalized in contrast
to English, for example. A comparison with English would
have to be carried out on the basis of the same stimuli,
however. We thus take first steps in exploring the influence
of aspect marking on event duration in Dutch.

Aims of the present study
In the present study, we draw a distinction between the
‘inherent’ duration of an event (i.e., the infinite and
unspecified time interval or duration of an event, as
expressed by bare (infinite) verb phrases, for example ‘to
write a paper’), which relates to world knowledge about the
normal course of an event, and the finite expression of event
duration by means of finite sentences or verb phrases,
relating to a specific situation (‘finite’ event duration).
Finite expressions of event duration can include a verbal
marker of progressive aspect, or not.
We ask whether speakers of Dutch perceive the duration
of an event differently, depending on the specific type of
verb form used (progressive or non-progressive) in a
sentence context. An example is ‘Wij zijn een artikel aan
het schrijven’ (lit.: we are a paper at-the-write; ‘we are
writing a paper’, progressive verb) versus ‘Wij schrijven een
artikel’ (we write a paper, non-progressive verb form). In
Dutch, both instances relate to a specific event, taking place
in the here and now.
Dutch speakers estimate the duration of events of
different types, described in written sentences, by dragging
a slider across a computer screen, using the mouse. Previous
studies show how performance on a spatial task may
accurately capture speakers’ conceptions of temporal
dimensions, such as duration (Casasanto & Boroditsky,
2008). Event sentences will be presented twice, once in a
progressive and once in a non-progressive condition. To
prevent participants from memorizing the estimated
duration of an event, as each is repeated, participants will
estimate duration in the absence of a concrete time scale.
Sentences will be presented in pairs, which remain the same
in both conditions, meaning that the ‘pair partner’ of an
event is thus the main point of reference for duration
estimation, rather than an absolute time line. Estimations in
minutes/seconds may be more susceptible to memory
effects, and may overrule subtle effects of (non-) aspect
marking.
We aim to find out how aspect interacts with the ‘natural’
or inherent duration of events, as judged by speakers on bare
verb phrases describing actions and events.
Experiment 1: ‘Inherent’ event duration
In experiment 1 native speakers of Dutch were asked to rate
all kinds of everyday events and actions described by bare
(non-finite) verb phrases (e.g., ‘to walk the dog’). Three
different samples were asked to rate their familiarity with
the events, in how far they are imaginable (to what extent is
the event likely to occur in the real world?) and the inherent
duration of the events or actions. All ratings were carried
out on a five-point scale.
Method
Participants In total, 30 native speakers of Dutch took part
in the experiment, consisting of three parts. They were
(PhD) students and postdoctoral researchers at Radboud
University Nijmegen (age range 19-35, balanced for
gender).
Materials Stimuli used were written infinite action phrases
(bare VPs) relating to everyday actions and events, and
described with infinite verb phrases, e.g., ‘to peel an apple’,
‘to open a can’, ‘to watch a football game’. Sentences were
placed in an online questionnaire in a randomized order, and
speakers were asked to give online ratings, and specify their
age and gender. In total, there were 150 different
events/actions.
Procedure Three different samples of 10 native speakers of
Dutch took part in three different short experiments,
designed as web questionnaires. First of all, the infinite
action phrases were rated for familiarity (‘how familiar are
you with this type of action?’) on a scale from 1 (highly
unfamiliar) to 5 (highly familiar). Only highly familiar
events were selected (ratings of 4 and 5) for Experiment 2.
A second sample rated the phrases as to what extent the
action was imaginable (rating 1: not imaginable at all, rating
5: highly imaginable). Furthermore, another sample of 10
speakers rated the duration of the infinite action phrases in
relation to a ‘standardized’ event, i.e., to boil pasta, which
was specified as lasting for about 7-8 minutes (rating 1:
much shorter than boiling pasta, rating 5: much longer than
boiling pasta). This latter rating was conducted to ensure
homogeneity of inherent event duration estimations.
Results
The three rating tasks in Experiment 1 resulted in the
selection of 78 different events. All other items were
discarded due to a low degree of familiarity, the fact that
they were not imaginable, or whether duration ratings
showed a high degree of heterogeneity. All in all, 72 items
were discarded. The 78 events were divided into three
categories of inherent event duration (26 items in each
category), on the basis of the duration ratings obtained:
short (e.g., ‘to turn a key’, ‘to light a candle’), medium (e.g.,
‘to set the table’; ‘to polish a shoe’) and long (e.g., ‘to watch
a dvd’, ‘to wash a car’). Items with an average rating of
between 1 and 2 were characterized as ‘short’ events (range
of ratings: 1 – 1.67). Items with an average rating of
between 4 and 5 were classified as ‘long’ events (range of
ratings: 4.11 – 5). Medium events were items with an
average rating of between 1.67 and 4.11.
The 78 items with homogeneously-rated inherent event
duration, categorized in three groups (short, medium and
long), were used as materials for Experiment 2.
Experiment 2: 'Finite' event duration
In Experiment 2 we asked native speakers of Dutch to
estimate the duration of events, as described in whole
sentences, marked with or without progressive aspect.

Method
Participants In the present study 27 native speakers of
Dutch took part, who were all students at Radboud
University (age range: 18-32, 16 female, 11 male), and did
not have an advanced level of proficiency in a second (or
third) language. This was established on the basis of their
answers in a language background questionnaire. Students,
who reported a stay of over three months in a foreign
language country, were excluded from participation.
Materials Stimuli consisted of written sentences describing
everyday situations and events. There were in total 78 items,
describing 78 different events. Each item was used for a
progressive and non-progressive sentence and paired with
an item with matching inherent event duration. There were
thus 13 pairs in each duration category (short-short pairs,
medium-medium pairs, long-long pairs). For the pairings,
care was taken to avoid any thematic or semantic
relatedness between the two items. Sentences were
presented as pairs to provide a kind of reference point for
the duration estimations, within each trial. Pairs were
always presented in the same aspect (either progressive or
non-progressive). The agents of all actions (the subjects of
the sentences) were described with two specific names,
‘Jan’ and ‘Paul’ in all cases.
Procedure Before subjects came to the lab, they were asked
to carry out the same online familiarity rating task as in
Experiment 1, dealing with all 78 bare event phrases.
Ratings were again made on a scale from 1 to 5. In the lab,
subjects were told that on each trial they would read two
sentences describing the situations in which two specific
persons, i.e., ‘Jan’ and ‘Paul’ were involved right now. They
were asked to imagine the situations of both Jan and Paul,
and to estimate how long the two agents would be engaged
in the activities described. Numbered sentences appeared
below each other on a computer screen in a centred position.
Within trials, sentences were of approximately the same
length, to avoid any visual bias. Lower down, two sliders
were presented and subjects were instructed to use the
mouse to drag the sliders from left to right, starting with the
top one, to estimate duration (Figure 2).
1. Jan is een sprookje aan het vertellen
2. Paul is de badkamer aan het poetsen
1.
2.
Figure 2: computer screen with sliders dragged slightly to
the right (progressive aspect condition, ‘long’ events: ‘John
is telling a fairytale’, ‘Paul is cleaning the bathroom’)
Subjects were instructed that the further they dragged the
mouse to the right, the longer they estimated the agent to be
engaged in the activity. Furthermore, it was explicitly stated
that if they dragged the slider to the right only slightly, this
would mean that Jan or Paul are engaged in the activity for a
very short time. If they dragged the slider to the utmost
right, this would mean that Jan or Paul are performing the
activity for a long period of time. The particular part of the
slider that was dragged, turned red. Subjects were able to
adjust their estimations. After estimating the duration of
both sentences, they proceeded to the next trial by clicking a
button.
In order to ensure that participants were actually aware of
the surface sentence structure, and did not only focus on the
bare event characteristics, a question relating to the contents
of one of the preceding sentences appeared randomly.
Subjects were asked to decide whether they had read that
sentence before, by clicking yes or no on a button box. The
question sentences were correct half of the time, and the
other half contained errors with regard to the type of object
described (e.g., for sentence 2 above: Paul is cleaning the
kitchen) or the type of aspect used (e.g., for sentence 2: Paul
cleans the bathroom). Each sentence pair appeared twice,
once in the non-progressive condition, and once in the
progressive condition. All trials were pseudo-randomized,
so that each repeated item appeared in the second half of the
experiment (the second set of 39 trials), to ensure enough
distance between repeated items. The occurrence of
progressive or non-progressive sentence pairs in the first or
second half of the experiment was varied between subjects.
After filling out a sociolinguistic questionnaire, subjects
were asked to estimate the precise duration of the different
events (described in bare VPs) in minutes (pencil-and-paper
test). This was done to double-check, whether the events
were rated as belonging to the same duration categories as
those established in Experiment 1.
Results
a) Familiarity ratings All 78 event phrases were rated as
familiar (4) to highly familiar (5), replicating the results
from experiment 1.
b) Online event duration estimations of sentences For the
analysis, we focused on the values of the x-coordinates on
the computer screen only, equalling the distance the mouse
was dragged towards the right side of the screen. We
analyzed our data using mixed effects models (R, lme4
package). Our goal was to fit a model that would explain the
estimations made by the subjects as the result of the impact
of various variables, i.e. fixed and random effects. Our fixed
effects were ‘inherent duration’ (‘dur’) (long, medium,
short) and ‘aspect’ (progressive, non-progressive). The
variables were coded as follows: for ‘dur’, the short event
category was coded as the base level (-1 short, 0 long, 1
medium) and for ‘aspect’ we coded the non-progressive
verb form as the base level (-1 non progressive form, 1 aan
het form). We also aimed at controlling further influences

caused by the experimental design, by taking into account
random effects in our model.
The random factors we originally considered were subject,
item, and pair. For ‘subject’ we included by-subject random
intercepts, as well as a by-subject random slope, which
allowed the predictions for ‘inherent duration’ to shift by a
fixed amount for each subject
1
. With respect to the random
factor ‘item’, two things are important. First, every item
(event) belongs to one and only one event duration category.
Item is thus a nested random factor. We incorporated this by
adding a variable which covered the ‘item:dur’
interrelationship; this term was also included as a random
factor. Second, subjects always rated event pairs and not
single events. The pairing of items remained fixed
throughout the experiment, for each subject. We thus did not
add pair as a separate random factor, as the nested
‘item:dur’ term would sufficiently capture the variance
stemming from random item selection. In general, we
follow an approach by Barr et al. (2012) in which the
authors argue for a maximal approach, that is, “valid
statistical inferences using LMEMs require maximal
random-effects structures wherever possible …”(p.1).
We log-transformed and centred all duration estimations
(see Footnote 1 for the formula in R).
Let us turn directly to the fixed effects section in our
model (Table 1 below: asp.1 is aan het condition; dur.0 is
long, dur.1 is medium event type)
Table 1: Fixed effects in the mixed model
Estimate
Std. Error
t value
p value
Intercept
-0.96918
0.10877
-8.910
dur.0
1.90106
0.12323
15.427
<.001**
dur.1
1.00654
0.09792
10.279
<.001**
asp.1
0.04819
0.02364
2.039
.041*
dur.0:asp.1
-0.07118
0.03343
-2.129
.033*
dur.1:asp.1
-0.06721
0.03341
-2.012
.044*
As predicted, for ‘inherent duration’ (‘dur’) we find high t-
values (long events t = 15.43; medium events t = 10.28),
showing that, in contrast to the base level (short events), the
two other event types are estimated as significantly longer.
There was a significant main effect of ‘aspect’ (p = .041)
2
,
meaning that short events were estimated as having a longer
duration in sentences marked with the aan het form, when
compared to the same events described with non-
progressive verbs. Looking at the interaction effects, we
find that, compared to our base level, medium and long
events marked with the progressive form are estimated as
significantly shorter (both p values > .05).
1
The formula in R was the following:
scalest ~ 1 + dur * asp + (1+dur | subject)+(1 | nes.item)).
2
We calculated p values on the basis of the t-values, using the
following code in R:
tvalues <- fixef(model) / sqrt(diag(vcov(model)))
pvalues <- 2*(1-pnorm(abs(tvalues)))
To exclude the possibility that the above pattern of results is
due to the presence of outliers, 32 extreme values (.008 %),
with a standardized residual at a distance greater than 3
standard deviations from zero, were removed from the data,
and the model was refitted.
Table 2: Fixed effects in the mixed model on trimmed data
Estimate
Std. Error
t value
p value
Intercept
-0.97638
0.10859
-8.991
dur.0
1.93245
0.12593
15.345
<.001**
dur.1
1.03100
0.09883
10.432
<.001**
asp.1
0.04781
0.02157
2.217
.027*
dur.0:asp.1
-0.06942
0.03056
-2.272
.023*
dur.1:asp.1
-0.08131
0.03052
-2.665
.008*
The trimmed model (Table 2) shows the same significant
results for the predictors and their interactions. We conclude
that the statistical inferences made in the original model are
not confounded by extreme values in the data set.
c) Inherent event duration estimations (bare VPs) Table
3 below displays the average and SD of the duration
estimations for the infinite event phrases; these estimations
were carried out after the actual experiment. The numbers
displayed are duration estimations in minutes.
Table 3: Inherent duration estimations, in minutes
Short
Medium
Long
Average
2.25
11.20
80.24
SD
1.52
11.27
50.52
Lower
1
2
14.50
Upper
6.01
58.67
206.38
The absolute duration estimations support the division into
the three categories of inherent event duration, based on
Experiment 1.
General discussion
In Experiment 1, we established three categories of highly
familiar, everyday events of different ‘inherent’ duration
(short, medium, long events), on the basis of three rating
tasks. In Experiment 2, we used those items and specifically
assessed the effect of aspect marking on subjects’ duration
estimations of the three event types, by means of the ‘drag-
the-slider-technique’.
First of all, with respect to the different ‘inherent’ event
duration categories, the findings indicate that the method is
valid; medium and long events were estimated as lasting
significantly longer than short events. The duration
estimations made by subjects using a slider on the computer
screen, without a fixed time scale, reflect the time spans
which are inherently part of the conceptual representation of
events, showing that spatial tasks are informative about
people’s thinking about time (Casasanto & Boroditsky,
2008).

Second, we find a significant interaction between aspect
marking and inherent event duration, suggesting that aspect
affects the perceived duration of events described in
sentences in a specific way: In Dutch, short events are
estimated as having a longer duration when described in the
progressive aan het form, whereas medium and long events
are estimated as having a shorter duration, when compared
to estimations of the same events described by non-
progressive verb forms.
The mechanism underlying the patterns found is
explained by a time-relational analysis of aspect: As
described above (Klein 1994), progressive aspect marks that
the time span being talked about (TT) is placed within the
total event time (TSit) whereas unmarked (non-progressive)
verb forms are unspecific in this regard. With the
progressive, an internal time span is focused and explicitly
viewed as ‘in progress’. Short events inherently have a short
TSit, which can include a transition phase or change in state
(‘to open a bottle’: from ‘not open bottle’ to ‘open bottle’).
If language users describe such an event with progressive
aspect, the time span at issue is located within the event
time (Tsit), and attention is thus directed to the transition
phase. Language users experience this as stretching and
prolonging the duration of the event in their mental model.
For medium/long events, the temporal boundaries
(beginning and end) lie further apart (TSit is longer). There
are also phases with changes of state with the event ‘to
repair a bicycle’, for example, but it will typically have
longer duration. When events are described with progressive
verbs, attention is directed to a specific time interval that lies
in between the beginning and end of TSit, and, crucially, it
does not extend over the entire event. The duration of the
event will thus be perceived as shorter, compared to the
total time span for the entire event (TSit), as expressed in
non-progressive sentences.
World-knowledge about a specific event seems to play a
role for the interpretation of aspect – and both layers of
duration interact in our subjects’ mental models of the
events. In general, we provide further evidence that
grammatical aspect influences people’s representations of
events or situations (e.g., Anderson et al., 2008; Madden &
Zwaan, 2003).
Conclusions
In this study we investigated in how far grammatical aspect
has an influence on how people mentally represent the
duration of everyday, highly familiar events, described in
Dutch sentences. We distinguish between two ‘layers’ of
event duration, which are packaged together in sentences,
and which both contribute to the perceived duration of an
event. The first ‘layer’ consists of the ‘inherent’ duration of
an event, which is based on world knowledge. The second
layer consists of ‘finite’ temporal information, expressed by
tense and aspect. Our results imply that the inherent
duration of events is shared among speakers of a
language/culture. This inherent event duration is modulated
by grammatical aspect (aan het in Dutch; previous studies
show this for the –ing form in English).
We find psycholinguistic evidence for the function of
grammatical aspect in Dutch. By means of progressive
aspect, speakers take an ‘inside’ perspective on an event, by
selecting a time interval that falls within the total time
period of the event - leading to a complex interaction
between aspect marking and the inherent duration of events.
Acknowledgements
This study was funded by NWO Veni-grant 016.124.107 (to
M. Flecken). We thank Mary Carroll for proofreading and
commenting on an earlier version of this paper.
References
Anderson, S., Matlock, T., Fausey, C., & Spivey, M. (2008).
On the path to understanding on-line processing of
grammatical aspect. In B. Love, K. McRae, V. Sloutsky
(Eds.), Proceedings of the 30th Annual Conference
of the Cognitive Science Society (pp. 143–148). Mahwah,
NJ: Lawrence Erlbaum.
Barr, D., Levy, R., Scheepers, C. & Tily, H. (2013).
Random effects structure in mixed-effects models: Keep
it maximal. Journal of Memory and Language, 68, 255-
278.
Bergen, B., & Wheeler, K. (2010). Grammatical aspect and
mental simulation. Brain & Language, 112, 150-158.
Casasanto, D. & Boroditsky, L. (2008). Time in the Mind:
Using space to think about time. Cognition, 106, 579-593
Comrie, B. (1976). Aspect: An introduction to the study of
verbal aspect and related problems. Cambridge: CUP.
Dahl, Ø. (Ed.) (2000). Tense and Aspect in the Languages of
Europe. Berlin: Mouton de Gruyter.
Flecken, M. (2011a). Event conceptualization by early
bilinguals. Insights from linguistic and eye tracking data.
Bilingualism: Language & Cognition, 14 (1), 61-77.
Flecken, M. (2011b). What native speaker judgements tell
us about the grammaticalization of a progressive
aspectual marker in Dutch. Linguistics, 49 (3), 479-524.
Klein, W. (1994). Time in language. London: Routledge.
Madden, C. & Zwaan, R. (2003). How does verb aspect
constrain event representations? Memory & Cognition,
31, 663-672.
Magliano, J. & Schleich, M. (2000). Verb aspect and
situation models. Discourse processes, 29 (2), 83-112.
Stutterheim, C. v, Carroll, M. & Klein, W. (2009). New
perspectives in analyzing aspectual distinctions across
languages. In W. Klein & P. Li (Eds.), The expression of
time (pp. 195-216). Berlin: Mouton de Gruyter.