exploring individual differences in recognizing idiomatic
TRANSCRIPT
Exploring individual differences in recognizing idiomatic expressions in context
Mesian Tilmatine1,2, Ferdy Hubers2, & Florian Hintz3*
1Free University Berlin, Berlin, DE
2Centre for Language Studies, Radboud University, Nijmegen, NL
3Max Planck Institute for Psycholinguistics, Nijmegen, NL
--------------------------In press at Journal of Cognition--------------------------
Running title: Idiomatic expressions in context
Keywords: idiomatic expressions; individual differences; self-paced reading
*Corresponding author:
Florian Hintz
Max Planck Institute for Psycholinguistics
P.O. Box 310
6500 AH Nijmegen
The Netherlands
The archived materials, analysis scripts, logfiles and results can be found here:
https://hdl.handle.net/1839/5005965b-f11f-4c7a-a82d-ad6f6b6e58d4
2
Abstract
Written language comprehension requires readers to integrate incoming information with stored
mental knowledge to construct meaning. Literally plausible idiomatic expressions can activate both
figurative and literal interpretations, which convey different meanings. Previous research has shown
that contexts biasing the figurative or literal interpretation of an idiom can facilitate its processing.
Moreover, there is evidence that processing of idiomatic expressions is subject to individual
differences in linguistic knowledge and cognitive-linguistic skills. It is therefore conceivable that
individuals vary in the extent to which they experience context-induced facilitation in processing
idiomatic expressions. To explore the interplay between reader-related variables and contextual
facilitation, we conducted a self-paced reading experiment. We recruited participants who had recently
completed a battery of 33 behavioural tests measuring individual differences in linguistic knowledge,
general cognitive skills and linguistic processing skills. In the present experiment, a subset of these
participants read idiomatic expressions that were either presented in isolation or preceded by a
figuratively or literally biasing context. We conducted analyses on the reading times of idiom-final
nouns and the word thereafter (spill-over region) across the three conditions, including participants’
scores from the individual differences battery. Our results showed no main effect of the preceding
context, but substantial variation between readers and variation in contextual facilitation. We
encourage interested researchers to exploit the present dataset for follow-up studies on individual
differences in idiom processing.
3
Introduction
To understand sentences and discourse properly, readers must know facts about the world and
the plausibility of a described situation. In some cases, retrieving and combining the meaning of
individual words is not sufficient to activate the meaning intended by a sentence or discourse. That is,
there are cases where fixed sequences of words, also known as instances of formulaic language, carry
a meaning that does not emerge from its constituent words and that differ from the literal interpretation
of the word sequence (Abel, 2003).
A prominent type of fixed word sequences are idiomatic expressions (Wray & Perkins, 2000).
Previous research on idioms has shown that they are often processed faster than regular expressions,
because they are well-known, pre-established sequences of words that can be predicted (Tabossi,
Fanari & Wolf, 2009). An extensive debate in the field has concerned the mental representation and
processing of idioms. According to the ‘lexical representation hypothesis’, the meaning of idiomatic
expressions is represented as a single unit rather than being composed ‘on the fly’ (Bobrow & Bell,
1973; Swinney & Cutler, 1979; Gibbs, 1980). In contrast, compositional approaches assume that each
constituent word contributes to the meaning of an idiomatic expression (Gibbs & O’Brien, 1990;
Cacciari & Glucksberg, 1991, McGlone, Glucksberg, & Cacciari, 1994). Finally, hybrid models (e.g.,
Titone and Connine, 1999), representing a mixture of lexical representation and compositional
accounts, have received a lot of empirical support and are nowadays widely accepted. Such models
assume that there are external forces (e.g., idiom frequency, discourse context, language user
characteristics) that act upon the precise nature of an idiom’s meaning activation. A good test case for
examining the predictions of idiom processing and storage accounts are ‘literally plausible
expressions’.
Literally plausible idiomatic expressions, such as ‘to play with fire’, are cases where both the
literal (playing with fire) and the figurative (taking a risk) meaning of the expression are frequently
used. It has been shown that in order to select the intended meaning from the two alternatives, readers
make use of context, which may bias either the figurative or the literal interpretation (Cacciari &
Tabossi, 1988; Holsinger, 2013).
4
Beck and Weber (2020) conducted a self-paced reading study to investigate the effects of
context on the processing of idiomatic expressions. Their participants read idioms embedded in
sentences that varied in how literally plausible the idiom was and in whether the preceding context
was figuratively or literally biasing. The idioms were followed by a resolution phrase that indicated
whether the intended reading was figurative or literal. An example item with consistent resolution
phrases is given in (1): the biasing context is in italics, the idiom is bold font, and the resolution phrase
is underlined. Their results showed that both types of context facilitated processing when the
resolution phrase was consistent with the intended figurative or literal interpretation as compared to
when it was inconsistent. However, contexts biasing a literal interpretation facilitated processing only
in idioms that had a high potential for a literal interpretation.
(1) a. The fearless climber, who was on a climb alone in the mountains, was ready to play with
fire with any risk if necessary later on.
b. The young camper, who was already bored without any of his friends, was ready to play
with fire from the grill if necessary later on.
Individuals vary substantially in their ability to use language (Dąbrowska, 2018; Kidd et al.,
2018). A recent report demonstrated that fluid and crystallized and intelligence predicted the
comprehension of metaphors (Stamenković, Ichien, & Holyoak, 2019), a form of figurative language.
It is therefore conceivable that language users also differ in their ability to process idiomatic
expressions.
In a first step towards exploring skills that underlie individual differences in idiom processing,
Cacciari, Corrardini, and Ferlazzo (2018) conducted a cross-modal priming experiment. Their
participants heard idioms embedded in a sentence context that biased the figurative interpretation of
the idiom. Following auditory presentation, a written target word that was semantically related to the
idiom appeared on the screen. Participants performed a lexical decision task on the written word. The
underlying assumption was that participants who recognize an idiom quickly respond faster to the
semantically related target word due to spreading activation. Cacciari and colleagues tested whether
5
variability in lexical decision times could be explained by measures of participants’ non-verbal
processing speed, inhibitory control, working memory, cognitive flexibility, crystallized and fluid
intelligence, and personality traits. In terms of linguistic and general cognitive skills, their analyses
showed positive effects of working memory, inhibitory control and crystallized verbal intelligence on
lexical decision times (i.e., reflected in shorter RTs).
In sum, previous research has shown that processing is facilitated when idioms are embedded
in contexts biasing either a figurative or a literal interpretation—in the latter case, only if the idiom is
literally plausible. Moreover, there is evidence from a cross-modal priming paradigm for individual
differences in idiom processing, which have been related to differences in linguistic and general
cognitive skills.
One open question resulting from this body of research concerns the extent to which
contextual facilitation in idiom processing is subject to individual differences. That is, given the results
by Beck and Weber (2020) demonstrating facilitatory effects of context on self-paced reading times as
well as the individual differences data by Cacciari et al. (2018), it is likely that there is considerable
individual variation in how figuratively and literally biasing contexts affect readers’ processing of
idiomatic expressions. The present study addressed this question. We ran a self-paced reading
experiment via the internet using Dutch idioms selected from the normative idiom database by Hubers
et al. (2018, 2019). Next to their figurative meaning, all idioms had a high potential for being
interpreted literally. The idioms were embedded in short sentences. Our analyses focused on the
reading times of the idiom-final noun, the most meaning-bearing element in the fixed expression (see
Rommers et al., 2013; e.g., the word ‘fire’ in (1)). To allow for analyses of spill-over effects (Mitchell,
1994), we added a neutral adverb to follow the idiom-final noun, which marked the end of the
sentence. Participants read the sentences word by word in a non-cumulative, stationary window, self-
paced fashion. Importantly, each participant read each idiom in all of the three conditions: without a
preceding context (to assess the baseline reading time), or preceded by either a figuratively or literally
biasing context. This within-participants manipulation enabled us to determine for each participant to
what extent context affected their reading of the idiom. Moreover, since the experiment was conducted
via the internet, we expected large variation between participants pertaining to the speed of their
6
internet connection and the quality of their hardware (e.g., keyboard polling rate). We reasoned that a
within-participants design would mitigate these sources of noise as hardware-related noise should be
constant across conditions. On the other hand, repeating the same idiom twice within a participant –
albeit that the order of conditions was counterbalanced across lists – might affect their processing of
the idiom. We therefore offer two analyses—one based on the first encounter an idiom (in one of the
three conditions), and one based on the full dataset (including the two item reputations, Appendix C).
In general, we want to stress that the primary goal of this data report is to provide a brief motivation
and sample analysis for the present data. Interested researchers may further exploit the dataset for
targeted and/or exploratory analyses.
Our participants were native speakers of Dutch, who had recently taken part in a large-scale
individual-differences study where they completed 33 tests measuring linguistic and general cognitive
skills (Hintz et al., 2020). Hintz et al. (2020) used a latent-variable approach with multiple tests
tapping into the same cognitive construct. For the present analyses, we selected 19 of the 33 tests1 that
appeared relevant in the context of present study (cf. Cacciari et al., 2018). Specifically, the selected
tests tapped into five cognitive constructs: (1) Linguistic experience, (2) Non-verbal processing speed,
(3) Visual working memory, (3) Non-verbal intelligence, (4) Word reading skills, and (5) Predictive
sentence comprehension skills. We used principal component analysis to derive one score for each of
the 112 participants and each construct to be used in the analyses predicting idiom-final word and
spill-over reading times.
We predicted that compared to the condition where idioms were read in isolation, figuratively
and literally biasing contexts should lead to faster reading of idiom-final and spill-over words (Beck &
Weber, 2020). The crucial question was if and how individual differences in linguistic and general
cognitive skills affect idiom processing in context. While readers with higher levels of non-verbal
processing speed, non-verbal intelligence, and word reading skills may have a general processing
advantage (affecting reading times in all three conditions) over readers with lower scores on these
tests, the influence of visual working memory may be restricted to both context conditions. That is,
1 As for the present data, all data collected by Hintz et al. (2020) are publicly available. Thus, if researchers
would like to include additional/different individual-differences variables, they may download the data from UK
Data Archive: https://reshare.ukdataservice.ac.uk/854399/.
7
readers with larger visual working memory capacity may remember and use preceding contexts more
efficiently than readers with lower capacities. Similarly, readers with extensive linguistic experience
(e.g., vocabulary size, reading frequency) may have encountered idioms in a variety of different
contexts and may thus be faster at processing idioms in contexts than readers with less linguistic
experience. Finally, readers with better prediction skills during sentence comprehension may be able
to exploit the preceding contexts more efficiently for generating predictions about upcoming idiom-
final words than readers with worse prediction skills.
Method
Principal component analyses of linguistic and general cognitive skills
Before running the present study, we conducted a principal component analysis (using SPSS, version
27) on the test scores provided by Hintz et al. (2020). That is, for each of the five constructs (linguistic
knowledge, processing speed, visual working memory, sentence comprehension and prediction skills,
word reading skills), we tested how strongly the tests assumed to measure a given construct loaded on
its factor and how much variance was explained (Table 1; see Appendix A for descriptive statistics
and reliability measures of each included test and Figure 1 for correlations between the predictor
variables). The scores from Raven’s advanced progressive matrices test served as the measure of non-
verbal intelligence. We selected ‘oblimin rotation’ and extracted regression-based factor scores.
8
Table 1: Linguistic and general cognitive constructs: Factor loadings and variance explained.
Construct n Expl. variance Included tests Loading
Linguistic knowledge 112 58%
Peabody picture vocabulary 0.84
Spelling 0.75
Dutch Author recognition 0.82
Idiom recognition 0.54
Prescriptive grammar 0.83
Processing speed 107 53%
Auditory simple reaction time 0.71
Auditory choice reaction time 0.83
Letter comparison 0.48
Visual simple reaction time 0.74
Visual choice reaction time 0.81
Visual working memory 106 30%
Corsi block clicking forward 0.82
Corsi block clicking backward 0.85
Sentence comprehension
and prediction skills
105 55%
Gender cue activation 0.91
Verb semantics activation 0.91
Word reading skills 99 40%
Klepel 0.75
One-minute 0.83
Maximal speech rate 0.63
Phonological verbal fluency 0.72
Participants
We contacted the same 112 native Dutch participants who had previously taken part in the
study by Hintz and colleagues (2020) and invited them to take part in the present study. Forty-three of
them replied and participated in the self-paced reading experiment. They were paid €6. All participants
gave informed consent prior to participation. The ethics board of the Faculty of Social Sciences at
Radboud University (Nijmegen, NL) provided ethical approval to conduct the study. Two of the
9
participants were excluded from further analyses (Data pre-processing and analysis section); the
remaining 41 participants were on average 22.88 years old (SD = 2.8, range = 18-29; 9 male).
Figure 1: Correlations between individual differences predictors.
Materials
We selected 25 idiomatic expressions from the Dutch normative database by Hubers et al.
(2018, 2019). We embedded the idiomatic expressions in a carrier sentence (see (2) for an example;
context in italics, idioms in bold, see Appendix B for all items). Note that the idiom-final noun never
occurred in sentence-final position to avoid strategic processing effects and to enable spill-over
analyses. This was achieved by adding a semantically neutral word to the sentences.
10
Each of the 25 target sentences was presented in three conditions: in isolation, preceded by a
figuratively biasing context, and preceded by a literally biasing context, amounting to 75 experimental
trials containing an idiom. The context sentences were taken from two previous Dutch studies (van
Wonderen, Hubers, & Dijkstra, in prep.; van Ginkel, 2019) or created anew:
(2) a. In deze boekenwinkel heb ik laatst dat mooie boek gevonden. Ik tikte hem op de kop toen.
Transl.: In this bookstore, I recently found that nice book. I made a good deal that time.
b. Die hond heeft laatst mijn schoenen kapotgebeten. Ik tikte hem op de kop toen.
Transl.: That dog recently bit my shoes to pieces. I tapped him on the head that time.
In addition to the experimental materials, we created 75 non-idiomatic filler items. Twenty-
five of which were preceded by a context. Thus, there was an even number of trials with and without
context in the experiment. Finally, we created 30 comprehension questions that followed 20% of the
experimental and filler trials, which were included to ensure that participants kept focus.
All 50 experimental and 25 filler trials with preceding contexts were tested for plausibility in a
rating study conducted via the internet (within the Pavlovia web environment, Peirce et al., 2019),
involving 56 Dutch native speakers who were paid €4 for participation. These participants did not take
part in the main experiment or the study by Hintz et al. (2020). Participants were asked to judge how
well the second sentence (e.g., containing the ambiguous idiom) followed-up on the figuratively or
literally biasing context (Dutch: ‘Hoe goed volgt de tweede zin op de eerste?’). They responded to the
question by selecting a number on a 7-point Likert scale (ranging from 1, not well at all, to 7, very
well). The mean plausibility rating for the fifty experimental context-target sentence pairs was 4.37
(SD = 1.29, range: 1.47 to 6.17). The 25 trials with figuratively biasing contexts had an average
plausibility rating of M = 5.24 (SD = 0.78, range: 3.17 to 6.17); average plausibility rating of the 25
literally biasing contexts was M = 3.50 (SD = 1.10, range: 1.74 to 6.00).
11
Procedure
The experiment was programmed in jsPsych (de Leeuw, 2015) within the Pavlovia web
environment (Peirce et al., 2019), and run via the internet in participants’ browser. We created an
experimental list by shuffling the 75 filler items with the list of 75 experimental items. The 75
experimental items consisted of the 25 sentences with ambiguous idiomatic expressions, presented in
the three context conditions. We pseudo-randomized the order of the 150 trials and controlled that
there were minimally 10 trials in between two versions of the same idiom. Finally, we created five
additional versions of that list by counter-balancing the order of context conditions for the
experimental trials. That is, the lists varied in the order in which the context versions of a given idiom
were presented (e.g., neutral first, figurative context second, literal context third). The participants
were assigned to one of the six experimental lists. They consented to taking part by ticking off a
designated box. Participants were instructed to read the sentences silently as fast as possible while still
being able to comprehend their contents.
Context sentences were presented for the participant to read in one instance. There was no
time limit. Participants initiated the presentation of the target sentence by pressing the enter key. After
an interval of 500 ms, a fixation cross appeared for 500 ms, followed by the presentation of the first
word. Participants advanced to the next word by pressing the space bar. Reaction time for each word
was calculated as the difference between word presentation and button press. Content questions were
presented immediately after the last word in a sentence. Participants responded to the question by
pressing the keys J (yes-response) and N (no-response). After thirty trials, participants could take a
break. The inter-trial interval was 500 ms.
Data pre-processing and analysis
We used R (version. 3.5.1, R Core Team, 2018), and the libraries lme4 (Bates et al., 2015),
lmerTest (Kuznetsova, Brockhoff, & Christensen, 2017), effects (Fox, 2003), ggplot2 (Wickham,
2016), and performance (Lüdecke et al., 2020), to pre-process and analyse the data. Two participants
were excluded, because their accuracy on the comprehension questions was lower than 75% (65% and
71%), while all other participants scored substantially higher (M = 94,77%; SD = 4.23). Data cleaning
12
for the remaining 41 participants was performed on the idiom-final nouns and spill-over words
separately. In line with previous studies (Marsden et al., 2018; Prasad & Linzen, 2019), we excluded
words with reading times shorter than 100 ms and larger than 2000 ms from further analysis. This led
to the exclusion of less than 1% of the data in both analyses. Note that the following analysis was
conducted on the basis of the first encounter of an idiom (in one of the three conditions). For an
analysis of the full dataset, we refer the reader to Appendix C.
In separate models, we analysed the cleaned reading times of the idiom-final and spill-over
words using linear mixed effects regression analyses. The reading times were log-transformed to
correct for a right skew in the data. The three-level factor context (no context, figuratively biasing
context, literally biasing context) was coded using simple contrast coding (UCLA Statistical
Consulting Group, 2011). With simple contrast coding, the reference level is always coded as −1/3,
and the level that it is compared to is coded as 2/3. This way of coding is similar to treatment contrast
coding, but has the advantage that the intercept corresponds to the grand mean instead of
corresponding to the mean of the reference level. Moreover, factors outside of interactions can be
interpreted as main effects. As continuous predictor variables, we included linguistic knowledge,
visual working memory, processing speed, non-verbal IQ, word reading skills, sentence
comprehension and prediction skills, as well as length (number of letters) and frequency (Keuleers et
al., 2010) of the idiom-final noun/spill-over word and the idiom's transparency rating (Hubers et al.,
2018, 2019). All participant-related and item-related predictors were mean-centred and standardized.
We included random intercepts for items and participants. Adding any type of random slopes to the
model resulted in overfit.
Results
The average reading times and standard deviations of the idiom-final nouns and spill-over
words per context are presented in Table 2. Differences in reading times between both contexts were
very small for both the idiom-final nouns and the spill-over words. Similarly, the differences between
both context conditions and the no-context condition were small.
13
Table 2: Average reading times and standard deviations (ms)
by context for the idiom final word and the spill-over word.
Context Idiom final noun Spill-over word
Mean SD Mean SD
None 364.93 178.44 418.97 209.46
Figuratively biasing 359.80 155.94 419.23 207.98
Literally biasing 356.45 155.28 414.28 189.95
The results of the linear mixed effects regression analysis on the reading times of the idiom-
final noun are presented in Table 3. This analysis revealed a significant main effect of Word reading as
well as two interactions: one between context and Visual working memory and one between context
and Processing speed. The main effect of Word reading was negative, suggesting that participants with
better word reading skills read the idiom-final nouns faster than participants with lower word reading
skills. The two interactions are visualized in Figures 2 and 3. Figure 2 suggests that participants with
high visual working memory capacity were slower at reading idiom-final nouns in the literally biasing
context compared to both neutral and figuratively biasing context conditions. Figure 3 suggests that
participants with high non-verbal processing speed abilities (i.e., lower RTs) were faster at reading
idiom-final nouns in the literally biasing context compared to the neutral condition.
14
Table 3: Idiom-final noun regression model with logged RTs as dependent variable (the no-
context condition as the reference category).
Fixed effects β (SE) t p
Intercept 2.4830 (0.0567) 43.768 < 0.001 ***
Fig. biasing context (FBC) 0.0003 (0.0068) 0.049 0.961
Lit. biasing context (LBC) -0.0015 (0.0068) -0.216 0.829
Linguistic knowledge -0.0403 (0.025) -1.610 0.117
FBC × Ling. knowledge 0.0089 (0.0086) 1.032 0.302
LBC × Ling. knowledge 0.0101 (0.0085) 1.188 0.235
Visual working memory (WM) -0.0183 (0.0241) -0.758 0.454
FBC × Visual WM 0.0045 (0.0083) 0.540 0.589
LBC × Visual WM 0.0181 (0.0082) 2.195 0.028 *
Processing speed 0.0201 (0.0232) 0.865 0.393
FBC × Processing speed -0.0048 (0.0078) -0.622 0.534
LBC × Processing speed -0.0155 (0.0078) -1.997 0.046 *
Non-verbal IQ 0.0516 (0.0293) 1.763 0.087 .
FBC × Non-verbal IQ -0.0056 (0.0101) -0.554 0.580
LBC × Non-verbal IQ -0.0078 (0.0101) -0.779 0.436
Word reading -0.0623 (0.0221) -2.821 0.008 **
FBC × Word reading 0.0001 (0.0076) 0.009 0.993
LBC × Word reading 0.0115 (0.0075) 1.527 0.127
Sentence compr. & pred. (SPC) -0.0076 (0.0244) -0.312 0.757
FBC × SPC -0.0014 (0.0082) -0.169 0.866
LBC × SPC -0.0020 (0.0082) -0.239 0.811
Idiom transparency -0.010019 (0.01) -1.196 0.245
Idiom final noun frequency 0.0023 (0.0115) 0.204 0.840
Idiom final noun length 0.0065 (0.0089) 0.733 0.472
Random effects Variance SD
Participant 0.0164 0.128
Item 0.0019 0.043
Residual 0.0078 0.088
15
Figure 2: The interaction between Context and Visual working memory for the idiom-final word. The
error bands represent the 95% confidence interval.
Figure 3: The interaction between Context and Visual working memory for the idiom-final word. The
error bands represent the 95% confidence interval.
The results of the linear mixed effects regression analysis of the spill-over word reading times
are presented in Table 4. As in the idiom-final noun analysis, a significant main effect of Word reading
16
was observed, indicating that participants with better word reading skills were faster at reading the
spill-over word than participants with poorer word reading skills. No main effect of context or any
interactions involving context were observed.
Data discussion
In contrast to our hypotheses, we observed no main effect of context. However, as one would
expect, we observed that individuals with better word reading abilities read idiom-final and spill-over
words faster (in all three context conditions) than individuals with poorer word-reading skills. These
effects were seen in both types of analyses we conducted—based on the first encounter of a given item
(main analysis) and based on the full dataset (see Appendix C). While the fact that better reading
ability led to overall faster reading of idiom-final and spill-over words is not necessarily a novel
finding, it does demonstrate that the present self-paced reading experiment (conducted via the internet)
indeed picked up individual differences as measured in a different study, which was conducted almost
one year before the present experiment (Hintz et al., 2020).
In the analysis based on the first encounter of an item, we additionally observed evidence for
modulatory influences of visual working memory and non-verbal processing speed on reading idiom-
final targets in the literally biasing context (but not on no-context and figuratively biasing context)
condition: Individuals with better processing speed abilities read idiom-final nouns in that condition
faster than individuals with worse processing speed abilities, relative to the neutral condition,.
Moreover, readers with higher visual working memory capacities had longer RTs for idiom-final
nouns than readers with lower capacities. These effects suggest that participants were differentially
affected by idioms presented in contexts that biased the literal interpretation of an idiom’s constituent
words. One possible linking hypothesis for this data pattern is that individuals with better processing
speed abilities might have been able to link the preceding (‘deidiomatizing’) context to the unfolding
target sentence more quickly than individuals with lower processing speed abilities could. They were
thus faster and more efficient at switching off the idiomatic meaning, which led to faster target
processing. The inhibitory effect of visual working memory on target word processing is in contrast to
our hypotheses, which predicted that readers with larger visual working memory capacity should
17
remember and use preceding contexts more efficiently than readers with lower capacities, leading to
faster target word processing. The opposite was the case and we cannot offer a good account for this
finding. Future users of the data resource could explore this finding in more detail and, for example,
conduct analyses where multiple individual-differences predictors (among others, visual working
memory) interact.
In general, from a statistics point of view, future research could explore different ways of
analyzing the data. One may, for example, fit regression models with different random-effect
structures than the one used in the present model. Similarly, further work could address whether and if
so, how, repeating the same idiom twice within participants affected their processing. That is, while
the main effects of Word reading were consistent across both types of analyses, the two interactions
involving the literally biasing context condition discussed above were not observed when item
repetitions were included. Instead, we saw effects of non-verbal IQ and linguistic knowledge (in
interaction with the literally biasing context condition; see Appendix C, for a more detailed
description).
In sum, the present data resource offers many exciting avenues for conducting additional
exploratory and/or targeted analyses, especially when linked to the dataset provided by Hintz et al.
(2020). We hope that researchers make use of it to advance the field of idiom processing and/or
individual differences. The data can be accessed at https://hdl.handle.net/1839/5005965b-f11f-4c7a-
a82d-ad6f6b6e58d4. Interested researchers need to create a free account with the Archive of the Max
Planck Institute for Psycholinguistics providing a user name, email address, their full name and
affiliation. Alternatively, in case their institution is part of one of the supported Identity Federations
(Shibboleth), which is the case for many academic/research institutions, interested individuals may
simply use their own institutional account to log in. Use of the data is confined to academic purposes.
18
Table 4: Spill-over word regression model with logged RTs as the dependent variable (with the
no-context condition as the reference category).
Fixed effects β (SE) t p
Intercept 2.5070 (0.1409) 17.794 < 0.001 ***
Fig. biasing context (FBC) 0.0035 (0.0086) 0.403 0.687
Lit. biasing context (LBC) 0.0031 (0.0086) 0.360 0.719
Linguistic knowledge -0.0400 (0.0244) -1.638 0.111
FBC × Ling. knowledge 0.0002 (0.0107) 0.023 0.982
LBC × Ling. knowledge 0.0097 (0.0107) 0.908 0.364
Visual working memory (WM) -0.0321 (0.0235) -1.364 0.182
FBC × Visual WM 0.0002 (0.0104) 0.016 0.987
LBC × Visual WM 0.0124 (0.0103) 1.203 0.229
Processing speed 0.0248 (0.0226) 1.098 0.280
FBC × Processing speed -0.0013 (0.0097) -0.138 0.890
LBC × Processing speed -0.0101 (0.0097) -1.044 0.297
Non-verbal IQ 0.0525 (0.0285) 1.840 0.075 .
FBC × Non-verbal IQ 0.0015 (0.0126) 0.117 0.907
LBC × Non-verbal IQ 0.0035 (0.0126) 0.274 0.784
Word reading -0.0591 (0.0215) -2.746 0.010 **
FBC × Word reading -0.0027 (0.0095) -0.281 0.779
LBC × Word reading -0.0048 (0.0095) -0.501 0.617
Sentence compr. & pred. (SPC) -0.0179 (0.0238) -0.754 0.456
FBC × SPC -0.0040 (0.0103) -0.390 0.697
LBC × SPC 0.0003 (0.0103) 0.025 0.980
Idiom transparency -0.0166 (0.0110) -1.508 0.147
Spill-over word frequency 0.0019 (0.0246) 0.077 0.939
Spill-over word length 0.0105 (0.0071) 1.492 0.151
Random effects Variance SD
Participant 0.0154 0.124
Item 0.0025 0.050
Residual 0.0121 0.110
19
Acknowledgements
This study was performed while MT was enrolled as a Master’s student at Radboud University in
Nijmegen (NL) in their Cognitive Neuroscience program. We thank the Donders Institute for Brain,
Cognition and Behaviour for providing Pavlovia software licenses to run the web experiments. In
addition, we are grateful to Ton Dijkstra for his comments on an earlier version of the manuscript. Part
of this work is based on the research program ‘Free Competition in the Humanities’ with project
number 360-70-510 NWO ISLA, which is financially supported by the Netherlands Organisation for
Scientific Research (NWO). MT was funded by a grant within the ‘Empirical Study of Literature
Training Network’, funded by European Union Horizon 2020 (Marie Skłodowska-Curie, grant no.
860516). FloH was funded by the Netherlands Organization for Scientific Research (NWO),
Gravitation grant ‘Language in Interaction’ (grant number 024.001.006).
20
References
Abel, B. (2003). English idioms in the first language and second language lexicon: A dual
representation approach. Second Language Research, 19(4), 329-358.
Bates, D., Maechler, M., Bolker, B., Walker , S. (2015). Fitting Linear Mixed-Effects Models Using
lme4. Journal of Statistical Software, 67(1), 1-48.
Beck, S. D., & Weber, A. (2020). Context and Literality in Idiom Processing: Evidence from Self-
Paced Reading. Journal of Psycholinguistic Research, 49(5), 837-863.
Bobrow, S. A., & Bell, S. M. (1973). On catching on to idiomatic expressions. Memory & Cognition,
1(3), 343-346. https://doi.org/10.3758/bf03198118
Cacciari, C., & Glucksberg, S. (1991). Chapter 9 understanding idiomatic expressions: The
contribution of word meanings. Advances in Psychology, 217-240.
https://doi.org/10.1016/s0166-4115(08)61535-6
Cacciari, C., Corrardini, P., & Ferlazzo, F. (2018). Cognitive and personality components underlying
spoken idiom comprehension in context. An exploratory study. Frontiers in Psychology, 9.
Cacciari, C., & Tabossi, P. (1988). The comprehension of idioms. Journal of Memory and Language,
27(6), 668-683.
Dąbrowska, E. (2018). Experience, aptitude and individual differences in native language ultimate
attainment. Cognition, 178, 222-235.
Fox, J. (2003). Effect Displays in R for Generalised Linear Models. Journal of Statistical Software,
8(15), 1-27.
Gibbs, R. W. (1980). Spilling the beans on understanding and memory for idioms in conversation.
Memory & Cognition, 8(2), 149-156. https://doi.org/10.3758/bf03213418
Gibbs, R. W., & O'Brien, J. E. (1990). Idioms and mental imagery: The metaphorical motivation for
idiomatic meaning. Cognition, 36(1), 35-68. https://doi.org/10.1016/0010-0277(90)90053-m
Hintz, F., Dijkhuis, M., van‘t Hoff, V., McQueen, J. M., & Meyer, A. S. (2020). A behavioural
dataset for studying individual differences in language skills. Scientific Data, 7(1), 1-18.
Holsinger, E. (2013). Representing idioms: Syntactic and contextual effects on idiom processing.
Language and Speech, 56(3), 373-394.
21
Hubers, F., Cucchiarini, C., Strik, H., & Dijkstra, T. (2019). Normative data of Dutch idiomatic
expressions: Subjective judgments you can bank on. Frontiers in Psychology, 10.
Hubers, F., van Ginkel, W., Cucchiarini, C., Strik, H., & Dijkstra, T. (2018). Normative data on
Dutch idiomatic expressions: Native speakers. DANS [Dataset].
https://doi.org/10.17026/dans-zjx-hnsk
Keuleers, E., Brysbaert, M., & New, B. (2010). SUBTLEX-NL: A new measure for Dutch word
frequency based on film subtitles. Behavior Research Methods, 42(3), 643-650.
Kidd, E., Donnelly, S., & Christiansen, M. H. (2018). Individual differences in language acquisition
and processing. Trends in Cognitive Sciences, 22(2), 154-169.
Kuznetsova, A., Brockhoff, P. B., Christensen, R.H.B. (2017). lmerTest Package: Tests in Linear
Mixed Effects Models. Journal of Statistical Software, 82(13), 1-26.
Lüdecke, D., Makowski, D., Waggoner, P. & Patil, I. (2020). Assessment of Regression Models
Performance. CRAN. Available from https://easystats.github.io/performance
McGlone, M. S., Glucksberg, S., & Cacciari, C. (1994). Semantic productivity and idiom
comprehension. Discourse Processes, 17(2), 167-190.
https://doi.org/10.1080/01638539409544865
Marsden, E. J., Thompson, S., & Plonsky, L. (2018). A Methodological Synthesis of Self-Paced
Reading in Second Language Research: Methodological synthesis of SPR tests. Applied
Psycholinguistics, 39(5), 861–904.
Mitchell, D.C. (1994). An evaluation of subject-paced reading tasks and other methods for
investigating immediate processes in reading. In D. E. Kieras &M. A. Just (Eds.), New
methods in reading comprehension research (pp. 69–90). Hillsdale: Erlbaum
Peirce, J., Gray, J. R., Simpson, S., MacAskill, M., Höchenberger, R., Sogo, H., ... & Lindeløv, J. K.
(2019). PsychoPy2: Experiments in behavior made easy. Behavior Research Methods, 51(1),
195-203.
Prasad, G., & Linzen, T. (2018). Do self-paced reading studies provide evidence for rapid syntactic
adaptation? PsyArXiv. https://osf.io/qd8ye/
22
R Core Team (2018). R: A language and environment for statistical computing. R Foundation for
Statistical Computing, Vienna, Austria.URL https://www.R-project.org/.
Rayner, K., & Pollatsek, A. (2013). Basic processes in reading. In D. Reisberg (Ed), Handbook of
Cognitive Psychology (442-461). Oxford, UK: Oxford University Press.
Rommers, J., Dijkstra, T., & Bastiaansen, M. (2013). Context-dependent semantic processing in the
human brain: Evidence from idiom comprehension. Journal of Cognitive Neuroscience,
25(5), 762-776.
Stamenković, D., Ichien, N., & Holyoak, K. J. (2019). Metaphor comprehension: An individual-
differences approach. Journal of Memory and Language, 105, 108-118.
https://doi.org/10.1016/j.jml.2018.12.003
Swinney, D. A., & Cutler, A. (1979). The access and processing of idiomatic expressions. Journal of
Verbal Learning and Verbal Behavior, 18(5), 523-534.
Tabossi, P., Fanari, R., & Wolf, K. (2009). Why are idioms recognized fast? Memory & Cognition,
37(4), 529-540.
Titone, D. A., & Connine, C. M. (1999). On the compositional and noncompositional nature of
idiomatic expressions. Journal of Pragmatics, 31(12), 1655-1674.
https://doi.org/10.1016/s0378-2166(99)00008-9
Van Ginkel, W. (2019). 'It's all smooth sailing': Figurative and literal aspects of language
comprehension (PhD Dissertation). Donders Series, Nijmegen, The Netherlands.
Van Wonderen, E., Hubers, F., & Dijkstra, T. (in prep.). Single Words in Idiom Processing:
Activation of Word Forms and Literal Word Meanings.
Wickham, H. (2016). ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag New York, 2016.
Wray, A., & Perkins, M. R. (2000). The functions of formulaic language: An integrated model.
Language & Communication, 20(1), 1-28.
23
Appendix A - Descriptive statistics and reliability measures of each individual-differences test
Table A1: Descriptive statistics and reliability measures taken from Hintz et al. (2020).
Domain Test N Mean (SD) Range Skewness a Kurtosis a Internal
consistency
Retest
reliability f
Errors
(%)
Outliers (%)
S1 S2
Linguistic
Knowledge
Peabody picture vocabulary test 112 56 (25) 0 – 95 -0.43 -0.79 0.96 b 0.91 - -
Spelling test 112 0.56 (0.18) 0.1 – 0.93 -0.43 -0.37 0.83 c 0.85 - -
Dutch author recognition test 112 0.2 (0.12) -0.03 – 0.6 0.62 0.47 0.93 c 0.95 - -
Idiom recognition test 112 0.76 (0.13) 0.4 – 1 -0.33 0.03 0.53 c 0.78 - -
Prescriptive grammar test 112 0.69 (0.13) 0.4 – 1 0.04 -0.65 0.74 c 0.86 - -
General
cognitive skills
Auditory simple RT test 112 Log: 2.35 (0.08)
Raw: 235 (48)
Log: 2.2 – 2.65
Raw: 160 – 459 -1.36 d 3.1d 0.9 de 0.59 d - 0.85 1.21 d
Auditory choice RT test 112 Log: 2.6 (0.09)
Raw: 417 (100)
Log: 2.41 – 2.86
Raw: 263 – 799 -0.6 d 0.15 d 0.96 de 0.76 d 3.75 0.81 0.49 d
Letter comparison test 107 Log: 3.02 (0.08)
Raw: 1167 (251)
Log: 2.86 – 3.28
Raw: 748 – 2044 0.65 d 0.47 d 0.89 de 0.83 d 6.89 2.01 0.08 d
Visual simple RT test 112 Log: 2.37 (0.05)
Raw: 244 (33)
Log: 2.24 – 2.55
Raw: 179 – 358 -0.54 d 0.51 d 0.86 de 0.58 d - 0.49 1.88 d
Visual choice RT test 112 Log: 2.62 (0.07)
Raw: 439 (90)
Log: 2.5 – 2.86
Raw: 321 – 822 0.88 d 0.73 d 0.95 de 0.78 d 4.13 0.19 0.49 d
Visual
working
memory
Corsi block clicking test forward
Corsi block clicking test backward
111
108
8 (2)
7 (2)
3 – 12
3 – 12
-0.08
-0.04
0.25
-0.15
0.53 c
0.71 c
0.39
0.49
-
-
-
-
Word reading
skills
Verbal fluency phonology 112 16 (4) 3 – 30 0.15 0.55 - 0.71 - -
Maximal speech rate 106 Log: 3.60 (0.09)
Raw: 4028 (854)
Log: 3.39 – 3.82
Raw: 2458 – 6650 -0.24 -0.07 - 0.88 - -
One-minute test 111 90 (14) 56 – 116 -0.12 -0.57 0.46 c 0.79 - -
Klepel test 111 63 (12) 34 – 107 0.26 0.6 0.88 c 0.88 - -
Sentence
comprehension
and prediction
skills
Gender cue activation during
sentence comprehension 105 -588 (655) -1674 – 940 0.45 -0.95 0.88 e 0.88 1.6 0.25 0.82
Verb semantics activation during
sentence comprehension 112 -742 (673) -1701 – 1041 0.62 -0.72 0.86 e 0.76 0.96 0.65 0.68
See Usage Notes section for missing values in column ‘N’.
Values in S1 and S2 columns indicate the percentage of trials replaced during Stage 1 (trimming) and Stage 2 (outlier replacement) in the pre-processing pipeline. a Calculated based on aggregated performance indicators. b Internal consistency was calculated as Guttman’s Lambda-2 coefficient. c Internal consistency was calculated by adjusting split-half (odd–even) correlations with the Spearman-Brown prophecy formula. d Calculated based on log-transformed values. e Internal consistency was calculated as intra-class correlation coefficient 2 using the ‘psychometric’ package in R. f Test-retest reliability was operationalized as two-tailed Pearson’s correlation between performance on test days 1 and 2
24
Appendix B – Materials
Experimental items
Figurative context Literal context Target SPR idiom
(both lit./fig. translation) Associated question Mean Rating of plausibility
Figurative pair Literal pair
De oplichter heeft ons
geld afgetroggeld.
The swindler took
money from us.
De tuinman heeft het gazon
net gezaaid, dus mochten
we er niet overheen lopen.
The gardener had just sown
the lawn, so we were not
allowed to walk on it.
Hij leidde ons om de tuin
gisteren.
He led us around the garden
yesterday.
He fooled us yesterday.
[only in lit. cond.]:
Leidde de huismeester ons
om de tuin? (nee)
Did the housekeeper lead
us around the garden? (no)
5.36 3.5
Herman kan niet tegen
het zien van bloed.
Herman cannot stand
the sight of blood.
Met eten kun je die parkiet
altijd lokken.
You can always lure the
parakeet with food.
Hij ging meteen van zijn stokje
de vorige keer.
He immediately went off his
pole last time.
He immediately fainted last
time.
[only in fig. cond.]:
Kan Herman tegen het zien
van bloed? (nee)
Can Herman stand the
sight of blood? (no)
6.16 3.46
De leraar was snel boos.
The teacher was quick
to anger.
De kaars was snel
opgebrand.
The candle burned out
quickly.
Hij had een kort lontje
blijkbaar.
It had a short fuse, apparently.
He was temperamental,
apparently.
[only in neutral cond.]:
Had hij een kort lontje? (ja)
Did he have a short fuse?
(yes)
5.94 4.13
Zijn humeur was erg
slecht.
His mood was very bad.
Zijn pantoffels waren te
groot.
His house shoes were too
large.
Hij schoot steeds uit zijn slof
vanochtend.
He kept falling out of his
slipper this morning.
[only in lit. cond.]:
Waren zijn pantoffels te
groot? (ja)
5.90 3.25
25
He kept going off the rails this
morning.
Were his slippers too big?
(yes)
Het meisje wilde haar
ouders niet vertellen
waarom ze na moest
blijven.
The girl did not want to
tell her parents why she
was kept in detention
after school.
Moeder repareerde de gaten
in de truien van haar
kinderen.
Mother repaired the holes
in her children's jumpers.
Ze speldde hun iets op de
mouw toen.
She pinned something on their
sleeve then.
She lied then.
3.65 2.83
Eva is de laatste tijd
depressief.
Eva has been depressed
lately.
Tijdens het spelen op de
boerderij viel een van de
kinderen naar beneden.
While playing at the farm,
one of the children fell
down.
Ze zit diep in de put nu.
She is deep in the pit now.
She is in a dark place now.
5.52 3.71
De agent was in gesprek
met de verdachte.
The officer was talking
to the suspect.
De tandarts controleerde de
pijnlijke kies van de man.
The dentist controlled the
man's painful tooth.
Hij voelde hem aan de tand
gisteren.
He checked his tooth
yesterday.
He questioned him yesterday.
[only in fig. cond.]:
Was de agent in gesprek
met zijn baas? (nee)
Was the officer talking to
his superior? (no)
5.39 2.88
In deze boekenwinkel
heb ik laatst dat ene
mooie boek gevonden.
De hond van mijn ouders
heeft laatst mijn schoenen
kapot gebeten.
Ik tikte hem op de kop toen.
I tapped him on the head at the
time.
[only in fig. cond.]:
Was het boek duur? (nee)
4.55 4.33
26
I found that one
beautiful book in this
bookstore the other day.
My parents' dog recently
bit my shoes off.
I made a good deal at the time. Was the book expensive?
(no)
Zijn moeder bemoeit
zich erg met de keuzes
van haar kinderen.
His mother is very
meddlesome in her
children's choices.
De peuter speelt tijdens het
ontbijt altijd met haar eten.
The toddler always plays
with her food during
breakfast.
Ze heeft een vinger in de pap
helaas.
She has a finger in the mush,
unfortunately.
She is intrusive, unfortunately.
[only in neutral cond.]:
Heeft ze een vinger in de
pap? (ja)
Does she have a finger in
the mush? (yes)
Is she intrusive? (yes)
4.16 2.71
Jesse doet totaal niet
zijn best op het werk.
Jesse is not doing his
best at work at all.
De kleuter heeft een vieze
gewoonte.
The toddler has a nasty
habit.
Hij eet uit zijn neus soms.
He eats out of his nose
sometimes.
He is being lazy sometimes.
5.03 5.33
Die handelaar verkocht
gestolen fietsen tegen
hoge prijzen.
That dealer sold stolen
bicycles at high prices.
Het was nog erg donker in
de woonkamer omdat Geert
het licht niet aangezet had.
It was still very dark in the
living room because Geert
had not turned on the light.
Hij liep tegen de lamp
uiteindelijk.
He eventually ran into the
lamp.
He eventually was caught.
[only in lit. cond.]:
Was het donker in de
woonkamer? (ja)
Was it dark in the living
room? (yes)
5.07 4.13
De man doet niet lang
over het drinken van een
fles wijn.
De generaal wilde dat zijn
zoon ook het leger in zou
gaan.
The general wanted his son
to join the army too.
Hij maakte hem soldaat zo snel
het kon.
He made a soldier out of him
as quickly as he could.
[only in fig. cond.]:
Doet de man lang over het
drinken van een fles wijn?
(nee)
5.16 3.96
27
The man does not take
long to drink a bottle of
wine.
He emptied/finished it as
quickly as he could.
Does the man take long to
drink a bottle of wine? (no)
In het gezin bleef de
zachtaardige man op de
achtergrond.
In the family, the gentle
man remained in the
background.
De man lag in zijn stoel te
slapen op de boot.
The man was asleep in his
seat on the boat.
Zijn vrouw stond aan het roer
op dat moment.
His wife was at the helm at the
time.
His wife wore the britches at
the time.
[only in neutral cond.]:
Stond zijn dochter aan het
roer? (nee)
Was his daughter at the
helm? (no)
4.45 6.00
Jan liet Corine vroeger
altijd schrikken.
Jan used to startle
Corine.
Roy had gisteren meel
gehaald omdat Lisa wilde
gaan bakken.
Roy had fetched flour
yesterday because Lisa
wanted to bake.
Zij gaf hem een koekje van
eigen deeg nu.
She gave him a cookie of own
dough now.
She gave him a taste of her
own medicine now.
6.04 2.23
Wonderkind Jeanne
leerde wel drie nieuwe
talen tegelijkertijd.
Child prodigy Jeanne
learned as many as
three new languages at
the same time.
Karin moest na het ongeluk
een beenamputatie
ondergaan.
Karin had to undergo a leg
amputation after the
accident.
Zij kreeg die onder de knie
afgelopen week.
She got it/them under her
knees.
She mastered it/them last
week.
4.75 2.19
De ballerina deed
auditie voor de hoofdrol.
The ballerina auditioned
for the lead role.
De vrouw verkocht veel
van haar oude spullen.
The woman sold many of
her old things.
Ze haalde alles uit de kast
vanmiddag.
She pulled everything out of
the cabinet.
6.04 3.77
28
She pulled out all the stops this
afternoon.
Omdat veel werknemers
moesten overwerken,
schakelde de directeur
extra personeel in.
As many employees had
to work overtime, the
director called in extra
staff.
De machinist van de oude
stoomtrein opende een
uitlaat, want ze gingen wel
erg snel.
The conductor of the old
steam train opened an
exhaust, for they were
going very fast.
Dat haalde wat druk van de
ketel gelukkig.
Fortunately, that took some of
the pressure off the cauldron.
Fortunately, that made things
a bit less intense.
6.16 3.77
De dader werd
uiteindelijk niet schuldig
bevonden, omdat zijn
vrouw hem hielp.
The perpetrator was
eventually found not
guilty because his wife
helped him.
De vrouw van de achtbaan
gaf aan hoe lang het
jongetje moest zijn om mee
te mogen.
The woman at the roller
coaster indicated how tall
the little boy had to be to be
allowed on.
Zij hield hem een hand boven
het hoofd helaas.
She held a hand over his head,
unfortunately.
She protected him,
unfortunately.
5.29 2.03
Stijn en Maartje
irriteerden elkaar al de
hele dag.
Stijn and Maartje had
been annoying each
other all day.
Bas en Els zijn
concurrenten voor de winst
tijdens de
hardloopwedstrijd.
Bas and Els are
competitors for the win in
the running competition.
Uiteindelijk ging zij als eerste
door het lint vanochtend.
In the end, she was the first to
go through the finishing
straight this morning.
In the end, she was the first to
lose her temper this morning.
5.17 2.87
De reizigers hadden
geen kaartje toen de
Moeder had te veel spullen
meegenomen naar het park.
Ze vielen door de mand helaas.
6.04 2.90
29
conducteur hen kwam
controleren.
The passengers had no
ticket when the
conductor came to check
them.
Mother had taken too many
things to the park.
Unfortunately, they fell
through the basket.
Unfortunately, they did not get
away with it.
Wendy heeft het
ontzettend druk.
Wendy is very busy.
Eefje wil graag groter
lijken dan ze is.
Eefje would like to look
bigger than she is.
Ze loopt op haar tenen sinds
vorige week.
She has been tiptoeing since
last week.
She has been stressed since
last week.
5.04 3.97
De directeur was niet
tevreden met het huidige
beleid.
The director was not
satisfied with the current
policy.
De piraat vergistte zich in
de kant van het schip waar
het anker hoorde.
The pirate mistook the side
of the ship where the
anchor belonged.
Hij gooide het over een andere
boeg gisteren.
He threw it over another
[ship] bow yesterday.
He changed course yesterday.
[only in lit. cond.]:
Vergistte de piraat zich?
(ja)
Was the pirate making a
mistake? (yes)
5.75 3.03
Coen had nooit
verwacht dat hij de baan
echt zou krijgen.
Coen never expected
that he would actually
get the job.
De piloot vloog duidelijk
lager dan normaal.
The pilot was clearly flying
lower than normal.
Hij was in de wolken die dag.
He was in the clouds that day.
He was on cloud nine that day.
3.17 5.94
Teun winkelde erg vaak.
Teun shopped a lot.
Jorick heeft een naar
ongeluk gehad.
Hij had een gat in zijn hand
vroeger.
5.63 2.90
30
Jorick had a bad accident. He used to have a hole in his
hand.
He used to spend too much
money.
Hein was al heel oud.
Hein was already very
old.
De klusjesman was erg
goed met metaal.
The handyman was very
good with metal.
Hij legde het loodje
eergisteren.
He laid the lead [pipe]
yesterday.
He died the day before
yesterday.
5.63 1.74
31
Filler items
Context sentence SPR sentence Associated question
Met mijn vrienden was ik kamperen in Spanje,
waar het weer heel aangenaam was.
I was camping with my friends in Spain, where the
weather was very pleasant.
We sliepen in de open lucht zonder tent.
We slept in the open without a tent.
Waren we kamperen in Spanje? (ja)
Were we camping in Spain? (yes)
Als je gaat picknicken op het strand moet je wel
goed opletten.
When you go on a picnic on the beach, you have
to be careful.
Anders zit je eten snel onder het zand.
Otherwise, your food will soon be covered in
sand.
Moet je opletten dat je eten nat wordt? (nee)
Do you have to be careful not to get your food
wet? (no)
Vanuit haar balkon had Maaike goed zicht op de
vechtpartij op straat.
From her balcony, Maaike had a good view of the
brawl in the street.
Daarom riep de politie haar op als getuige.
That is why the police called her as a witness.
Had Maaike zicht op een verkeersongeluk? (nee)
Did Maaike have a view on a traffic accident?
(no)
Scheiden is altijd lastig, zeker als er kinderen bij
betrokken zijn.
Divorce is always difficult, especially when
children are involved.
Vaak is er dan ruzie over de voogdij.
There are often fights over custody.
Is er vaak ruzie over wie er voor de kinderen
mag zorgen? (ja)
Do you often argue about who can take care of
the children? (yes)
Suzanne had al een paar maanden lang last van
haar kies toen ze naar de tandarts ging.
Suzanne had been suffering from a toothache for
a few months when she went to the dentist.
Ze kreeg een verdoving tegen de pijn.
She was given an anaesthetic against the pain.
Had Suzanne last van haar teen? (nee)
Was Suzanne bothered by her toe? (no)
Naar de bioscoop gaan is heel leuk.
Voor de film begint, haal ik altijd een grote bak
popcorn.
Haal ik altijd een grote bak popcorn? (ja)
32
Going to the cinema is great fun.
Before the film starts, I always get a big bowl of
popcorn.
Do I always get a big bowl of popcorn? (Yes)
Johan en Linda vierden onlangs hun 25-jarig
huwelijk.
Johan and Linda recently celebrated their 25th
wedding anniversary.
Dus nodigden ze de hele familie uit voor een
groot feest.
So they invited the whole family to a big party.
Nodigden Johan en Linda al hun vrienden uit?
(nee)
Did Johan and Linda invite all their friends?
(no)
Zelf podcasts opnemen is helemaal niet zo
goedkoop.
Recording your own podcasts isn't that cheap.
Je moet namelijk heel wat betalen voor een goede
microfoon.
After all, you have to pay a lot of money for a
good microphone.
Is zelf podcasts opnemen duurder dan verwacht?
(ja)
Is recording your own podcasts more expensive
than you expected? (Yes)
Ferdy zorgt ervoor dat hij elke ochtend stevig
ontbijt.
Ferdy makes sure he has a hearty breakfast every
morning.
Vaak is er nauwelijks nog plek op zijn bord.
Often there is hardly any room left on his plate.
Emma vindt de geur van sigaretten
verschrikkelijk.
Emma hates the smell of cigarettes.
De kleren van haar vriendin stinken altijd naar
rook.
Her friend's clothes always stink of smoke.
Tijdens een marathon is het belangrijk om goed te
blijven drinken.
During a marathon, it is important to keep
drinking well.
Toen de loper door de laatste bocht kwam gaf
iemand hem water.
When the runner came through the last bend,
someone gave him water.
Als ik op een lange reis ga, leen ik vaak wat
boeken bij de bibliotheek.
Soms lever ik ze niet op tijd in en krijg ik een
enorme boete.
33
When I go on a long trip, I often borrow some
books from the library.
Sometimes I don't return them on time and get a
huge fine.
Marjolijn vindt gokken heel leuk maar is niet
goed in kaartspelen.
Marjolijn likes gambling very much but she is not
good at playing cards.
Vorige week verloor ze veel geld tijdens het
pokeren.
Last week she lost a lot of money playing poker.
Ik ging vaak op zaterdagochtend vissen met mijn
vader.
I often went fishing on Saturday morning with my
father.
We namen altijd een emmer wormen mee als aas.
We always took a bucket of worms as bait.
Jagers moeten er op letten dat ze niet per ongeluk
elkaar neerschieten.
Hunters have to take care not to accidentally
shoot each other.
Daarom dragen ze vaak een oranje vest.
That's why they often wear an orange vest.
Petra is erg blij dat ze tijdens de schaatswedstrijd
derde geworden is.
Petra is very happy that she came third in the
speed skating competition.
Zij wilde zo graag een plek op het podium.
She so badly wanted a place on the podium.
Bij de NS zijn er momenteel erg vaak
werkzaamheden.
At the moment, there are a lot of works on the NS
[Dutch National Railway Company].
Tussen Arnhem en Den Bosch rijden nu alleen
maar bussen.
Between Arnhem and Den Bosch only buses are
running at the moment.
34
Studentenhuizen staan bekend om hun niet al te
hygiënische toiletten.
Student houses are known for their not very
hygienic toilets.
Als je die schoon wilt maken, kun je beter
handschoenen dragen.
If you want to clean them, you'd better wear
gloves.
Mijn ouders waren erg blij toen mijn zus eindelijk
ging trouwen.
My parents were very happy when my sister
finally got married.
Mijn moeder moest zelfs huilen tijdens de
ceremonie.
My mother even cried during the ceremony.
Afgelopen zaterdag heb ik een tweedehands
televisie gekocht, maar ik kan niet van zender
wisselen.
Last Saturday I bought a second-hand television,
but I can't change the channel.
Wat er niet bij zat was een afstandsbediening.
What it didn't come with was a remote control/
Een van de boksers viel uiteindelijk op de mat.
One of the boxers finally fell on the mat.
De scheidsrechter riep de ander uit tot winnaar.
The referee declared the other one the winner.
Reizen met het vliegtuig duurt tegenwoordig veel
langer, omdat iedereen bang is voor terrorisme.
Travelling by plane takes much longer nowadays,
because everyone is afraid of terrorism.
Op vliegvelden staat er altijd een lange rij voor de
douane.
At airports, there is always a long queue for
customs.
Op een boerderij wonen heeft ook wel nadelen.
Living on a farm also has its disadvantages.
Elke ochtend werd ik vroeg wakker door het
gekraai van de haan.
Every morning, I was woken up early by the
crowing of the rooster.
35
Heb je afgelopen zaterdag de voetbalwedstrijd
gezien?
Did you see the football match last Saturday?
Die spits scoorde echt een hele mooie goal.
That striker scored a really nice goal.
Joeri gaat in de ochtendspits soms op de fiets naar
zijn werk.
In the morning rush hour, Joeri sometimes goes to
work by bike.
Dat gaat vaak zelfs sneller dan met de auto.
That is often even faster than going by car.
Soms schrikken de kinderen wel van het gebrul
van de leeuw.
Sometimes the lion's roar scares the children.
Schrikken de kinderen van het gebrul van de
tijger? (nee)
Does the roaring of the tiger frighten the
children? (no)
Hij is me nog aardig wat geld schuldig.
He still owes me a lot of money.
Ben ik hem nog aardig wat geld verschuldigd?
(nee)
Do I owe him a lot of money? (no)
Hij heeft duidelijk nog niet veel ervaring.
He clearly doesn't have much experience yet.
Heeft deze medewerker weinig ervaring? (ja)
Does this employee have little experience? (yes)
Planten in huis maken het meteen gezelliger.
Plants in the house make it cosier.
Maken planten het huis gezelliger? (ja)
Do plants make the house cosier? (yes)
Op het einde word je wel beloond met een
fantastisch uitzicht.
At the end, you are rewarded with a fantastic
view.
Word je op het einde beloond met lekker eten?
(nee)
Are you rewarded with good food at the end?
(no)
36
Daar werkte ik namelijk achter de kassa.
The thing is, I worked there behind the cash
register.
Werkte ik achter de kassa? (ja)
Was I working behind the counter? (Yes)
Hij wil heel graag een grotere tafel.
He would like a bigger table.
Heeft Martijn meubels nodig voor zijn
slaapkamer? (nee)
Does Martin need furniture for his bedroom?
(no)
De vrouw leefde al jaren alleen.
The woman had been living alone for years.
Leefde de vrouw alleen? (ja)
Did the woman live alone? (Yes)
Hij is me nog aardig wat geld schuldig.
He still owes me a lot of money.
Ben ik hem nog aardig wat geld verschuldigd?
(nee)
Do I owe him a lot of money? (no)
Hij moest toch lachen uiteindelijk.
He had to laugh in the end.
Voor al haar documenten wilde ze een nieuwe
laptop.
She wanted a new laptop for all her documents.
Dat wordt waarschijnlijk een flink litteken.
That will probably be quite a scar.
37
Na een paar uur wilde zij nog steeds niet het
zwembad uit.
After a few hours, she still did not want to leave
the pool.
Uiteindelijk besloten we te stoppen om te vragen
naar de weg.
Finally, we decided to stop and ask for directions.
Vleermuizen vind ik erg fascinerende dieren.
I find bats very fascinating animals.
Gelukkig is hij nu weer gemaakt door de
reparateur.
Fortunately, it has been repaired by the
repairman.
Toen ze de top bereikten, werden ze overweldigd
door het uitzicht.
When they reached the top, they were
overwhelmed by the view.
Als ze klaar is, is ze helemaal nat van het zweet.
When she's finished, she's all wet with sweat
.
Dat is overigens altijd al zo geweest.
It has always been like that, by the way.
38
Buiten bouwden ze een grote sneeuwpop.
Outside they built a big snowman.
Zij willen graag een groter huis.
They would like a bigger house.
Om mij te feliciteren, stuurde ze me nog wel een
kaartje.
To congratulate me, they sent me a card.
Het is altijd moeilijk om een goed cadeau te
bedenken.
It is always difficult to think of a good gift.
Die paar weken rust hebben hem zeker goed
gedaan.
Those few weeks of rest have certainly done him
good.
Tijdens de Vierdaagse is er behoorlijk veel lawaai.
During the Four Days Marches there is a lot of
noise.
Met een theoretische opleiding vind je
tegenwoordig toch geen baan.
You can't find a job with a theoretical education
these days anyway.
39
Ik zou de hele week pizza kunnen eten.
I could eat pizza all week.
Zij had veel te lang niet meer geschaatst.
She had not skated for far too long.
Gelukkig zie je de vlek niet zo goed op mijn
donkere trui.
Luckily you can't see the stain so well on my dark
jumper.
Op zondagavond kijkt ze altijd naar dat
programma van Lubach.
She always watches that show of Lubach [Dutch
late night show host] on Sunday evening.
Laatst heb ik koeien zien lopen op straat.
The other day, I saw cows walking on the street.
Naast kleding zijn we ook op zoek naar mooie
handtassen.
Apart from clothes, we are also looking for
beautiful handbags.
Ze gaat echt heel graag zwemmen.
She really likes to go swimming.
Ik heb heel wat zaadjes geplant in mijn moestuin.
40
I planted a lot of seeds in my vegetable garden.
We hadden echt veel te veel gegeten.
We had really eaten too much.
Ik sta vaak een lange tijd doelloos naar mezelf te
staren in de spiegel.
I often spend a long time staring at myself
aimlessly in the mirror.
Hij kookt nu vijf keer per week.
He cooks five times a week now.
Ze was zelden zo gelukkig geweest.
She had rarely been so happy.
Hij werd gelukkig snel in veiligheid gebracht door
de badmeester.
Fortunately, he was quickly taken to safety by the
lifeguard.
Sindsdien draagt hij altijd een helm.
Since then, he always wears a helmet.
Ik had mijn hond het liefst meegenomen op reis.
I would have preferred to take my dog with me on
a trip.
41
Je kunt dit natuurlijk ook uitbesteden aan een
verhuisbedrijf.
Of course, you can also outsource this to a
removal company.
Ze pakte het vliegtuig vorige week.
She took the plane last week.
Ze vroeg de dokter om een nieuwe afspraak.
She asked the doctor for a new appointment.
Bij schrijven is het begin altijd het moeilijkst.
When writing, the beginning is always the
hardest.
Bach is mijn favoriete componist tegenwoordig.
Bach is my favourite composer these days.
Hij moest daar erg om lachen.
He had to laugh about it.
Waar mogelijk neemt ze liever de lift.
Whenever possible, she prefers to take the lift.
De kinderen renden weg zo snel als ze konden.
The children ran away as fast as they could.
42
Even later hoorde ze een verschrikkelijke knal.
A little later she heard a terribly loud bang.
Hij werkte namelijk in de kerk als pastoor.
He was working in the church as a priest.
43
Appendix C – Analysis based on full dataset (including item repetitions within a participant)
The average reading times and standard deviations of the idiom-final nouns and spill-over words per
context are presented in Table C1. Differences in reading times between contexts were very small for
both the idiom-final nouns and the spill-over words.
Table C1: Average reading times and standard deviations
(ms) by context for the idiom final word and the spill-over
word.
Context Idiom final noun Spill-over word
Mean SD Mean SD
None 304.73 143.55 355.39 182.74
Figuratively biasing 302.89 138.10 358.24 184.04
Literally biasing 301.91 129.62 355.23 170.72
Figure C1 presents the mean differences between the figuratively biasing context and the no
context condition (left panels) and the mean differences between the literally biasing context and the no
context condition (right panels) per participants. Positive values indicate contextual facilitation. The
figure shows that participants were differently affected by our context manipulation.
44
Figure C1: The mean differences between contexts per participant for both the idiom-final nouns (upper
panels) and the spill-over words (lower panels). Error bars represent SDs.
The results of the linear mixed effects regression analysis on the idiom-final noun are presented
in Table C2. This analysis revealed significant main effects of Non-verbal IQ and Word reading. The
effect of non-verbal IQ was positive, indicating that participants with a higher non-verbal IQ were slower
in reading the idiom-final noun than participants with a lower non-verbal IQ. In addition, we observed a
facilitatory effect of word reading: Participants with better word reading skills were faster in reading the
idiom-final noun than participants with poorer word reading skills. The effects of these participant-related
predictors were not modulated by context.
45
Table C2: Idiom-final noun regression model with logged RTs as dependent variable (the no-
context condition as the reference category).
Fixed effects β (SE) t p
Intercept 2.4455 (0.0324) 75.486 <0.001 ***
Fig. biasing context (FBC) -0.0013 (0.0048) -0.274 0.784
Lit. biasing context (LBC) -0.0008 (0.0048) -0.168 0.866
Linguistic knowledge -0.038 (0.0234) -1.627 0.113
FBC × Ling. knowledge 0.0059 (0.0059) 1.000 0.317
LBC × Ling. knowledge 0.0088 (0.0059) 1.508 0.132
Visual working memory (WM) -0.0157 (0.0225) -0.697 0.491
FBC × Visual WM -0.001 (0.0056) -0.172 0.863
LBC × Visual WM 0.0091 (0.0056) 1.621 0.105
Processing speed 0.0184 (0.0216) 0.853 0.4
FBC × Processing speed 0.0035 (0.0054) 0.64 0.522
LBC × Processing speed -0.0027 (0.0054) -0.49 0.624
Non-verbal IQ 0.0587 (0.0273) 2.148 0.039 *
FBC × Non-verbal IQ -0.0081 (0.0068) -1.189 0.235
LBC × Non-verbal IQ -0.0105 (0.0069) -1.527 0.127
Word reading -0.0607 (0.0206) -2.947 0.006 ***
FBC × Word reading 0.0031 (0.0052) 0.596 0.551
LBC × Word reading 0.005 (0.0052) 0.964 0.335
Sentence compr. & pred. (SPC) -0.0118 (0.0227) -0.519 0.607
FBC × SPC 0.0013 (0.0057) 0.233 0.816
LBC × SPC -0.0001 (0.0057) -0.014 0.989
Idiom transparency -0.0014 (0.0049) -0.289 0.775
Idiom final noun frequency 0.0028 (0.0057) 0.491 0.629
Idiom final noun length -0.0017 (0.0044) -0.384 0.705
Random effects Variance SD
Participant 0.0144 0.120
Item 0.0004 0.020
Residual 0.0114 0.107
46
The results of the linear mixed effects regression analysis of the spill-over word are presented in
Table C3. Similar to the idiom-final noun analysis, a significant main effect of word reading was
observed, indicating that participants with better word reading skills were faster at reading the spill-over
word than participants with poorer word reading skills. This effect was not modulated by context.
Furthermore, the analysis revealed a significant interaction effect between linguistic knowledge and
context (visualised in Figure C2). The effect of linguistic knowledge was significantly stronger for spill-
over words in the no-context condition, than in the literally biasing context condition. In comparison to
the figuratively biasing context condition, the effect of linguistic knowledge in the no-context condition
was only marginally significantly stronger.
In sum, we observed no main effect of context. However, as predicted, we observed that
individuals with better word reading abilities read idiom-final and spill-over words (in all three context
conditions) faster than individuals with poorer word-reading skills. Non-verbal intelligence had a negative
effect: In all three context conditions, individuals with higher scores read idiom-final words more slowly
than individual with lower scores. Finally, we observed an interaction between Context and Linguistic
knowledge such that spill-over words were read faster in the no-context condition (compared to the
literally biasing condition) by individuals with larger rather than smaller linguistic knowledge.
47
Table C3: Spill-over word regression model with logged RTs as the dependent variable (the no-
context condition as the reference category).
Fixed effects β (SE) t p
Intercept 2.519 (0.101) 24.943 <0.001 ***
Fig. biasing context (FBC) 0.0039 (0.0058) 0.669 0.504
Lit. biasing context (LBC) 0.0033 (0.0058) 0.574 0.566
Linguistic knowledge -0.0341 (0.0251) -1.361 0.183
FBC × Ling. knowledge 0.0131 (0.0071) 1.849 0.065 .
LBC × Ling. knowledge 0.0165 (0.0071) 2.327 0.02 *
Visual working memory (WM) -0.0258 (0.0241) -1.069 0.293
FBC × Visual WM 0.0021 (0.0068) 0.314 0.753
LBC × Visual WM 0.0045 (0.0068) 0.662 0.508
Processing speed 0.0236 (0.0232) 1.018 0.316
FBC × Processing speed -0.0042 (0.0066) -0.64 0.522
LBC × Processing speed -0.0041 (0.0066) -0.632 0.527
Non-verbal IQ 0.0569 (0.0293) 1.942 0.061 .
FBC × Non-verbal IQ -0.0038 (0.0083) -0.453 0.651
LBC × Non-verbal IQ -0.0006 (0.0083) -0.077 0.939
Word reading -0.0599 (0.0221) -2.707 0.011 *
FBC × Word reading 0.0025 (0.0063) 0.394 0.694
LBC × Word reading 0.0011 (0.0063) 0.178 0.859
Sentence compr. & pred. (SPC) -0.0197 (0.0244) -0.809 0.424
FBC × SPC -0.0055 (0.0069) -0.793 0.428
LBC × SPC -0.0034 (0.0069) -0.496 0.62
Idiom transparency 0.001 (0.0078) 0.132 0.897
Spill-over word frequency -0.0064 (0.0174) -0.368 0.716
Spill-over word length 0.0024 (0.005) 0.481 0.635
Random effects Variance SD
Participant 0.0165 0.129
Item 0.0013 0.036
Residual 0.0167 0.129