ArticlePDF Available

How Listening to Music Affects Reading: Evidence From Eye Tracking


Abstract and Figures

The current research looked at how listening to music affects eye movements when college students read natural passages for comprehension. Two studies found that effects of music depend on both frequency of the word and dynamics of the music. Study 1 showed that lexical and linguistic features of the text remained highly robust predictors of looking times, even in the music condition. However, under music exposure, (a) readers produced more rereading, and (b) gaze duration on words with very low frequency were less predicted by word length, suggesting disrupted sublexical processing. Study 2 showed that these effects were exacerbated for a short period as soon as a new song came into play. Our results suggested that word recognition generally stayed on track despite music exposure and that extensive rereading can, to some extent, compensate for disruption. However, an irrelevant auditory signal may impair sublexical processing of low-frequency words during first-pass reading, especially when the auditory signal changes dramatically. These eye movement patterns are different from those observed in some other scenarios in which reading comprehension is impaired, including mindless reading.
Content may be subject to copyright.
Word count: 9679
How listening to music affects reading: Evidence from eye tracking
Han Zhang*, Kevin Miller, Raymond Cleveland, Kai Cortina
Combined Program in Education and Psychology
University of Michigan, Ann Arbor, MI, 48109, USA
*Correspondence to: Han Zhang, Combined Program in Education and Psychology, University
of Michigan, 610 E. University Ave., Ann Arbor, MI 48109, USA
Phone: +1 734 680 6031, E-mail:
Manuscript (revised)
The current research looked at how listening to music affects eye movements when college
students read natural passages for comprehension. Two studies found that effects of music
depend on both frequency of the word and dynamics of the music. Study 1 showed that lexical
and linguistic features of the text remained highly robust predictors of looking times even in the
music condition. However, under music exposure, (1) readers produced more re-reading, (2)
gaze duration on words with very low frequency were less predicted by word length, suggesting
disrupted sublexical processing. Study 2 showed that these effects were exacerbated for a short
period as soon as a new song came into play. Our results suggested that word recognition
generally stayed on track despite music exposure and that extensive re-reading can to some
extent compensate for disruption. However, an irrelevant auditory signal may impair sublexical
processing of low frequency words during first-pass reading, especially when the auditory signal
changes dramatically. These eye movement patterns are different from those observed in some
other scenarios where reading comprehension is impaired, including mindless reading.
Key words: music, reading, eye tracking, distraction
How listening to music affects reading: Evidence from eye tracking
Music Impairs the Reading Process
People are often exposed to background music in daily life. Previous studies have shown
that listening to music impairs performance of cognitively-demanding tasks, although its effect
size depends on multiple factors including the task involved (Banbury, Macken, Tremblay, &
Jones, 2001; Boyle, 1996), the type of music played (Cassidy & Macdonald, 2009; Furnham &
Allass, 1999; Perham & Sykora, 2012), and characteristics of the person (Crawford & Strapp,
1994; Furnham & Bradley, 1997). One cognitive task that has been consistently demonstrated to
be susceptible to music interference is reading comprehension, as typically demonstrated by
contrasting music versus no music (Dalton & Behm, 2007; Kampfe, Sedlmeier, & Renkewitz,
2010). However, both reading and music are dynamic processes, as the reader goes through
words with various processing difficulties and listens to auditory signals whose acoustic features
such as tone, pitch, and amplitude are constantly changing. Therefore, one might reason that the
effect of music may depend on both materials being read and acoustic transitions in the music.
The current study employed eye-tracking as it enables us to study the process of reading
with music in a moment-to-moment fashion. Although music has been shown to be distracting,
little is known about exactly how listening to music affects eye movements during reading. Most
eye movement reading research has been conducted with skilled readers reading in silence, with
the assumption that readers give their full attention to the text. However, people are often
exposed to auditory distractions while studying and working, and many people even voluntarily
choose to listen to music. More than 80 years ago, Cantril and Allport (1935) found that 68% of
college students reported that they listen to the radio while studying. A recent study shows that
59% of the students listened to music during a three-hour study session, with 21% listening to
music for over 90% of the time (Calderwood, Ackerman, & Conklin, 2014). The question of how
listening to music affects reading eye movements is one that has practical implications for
helping students to study effectively as well as theoretical implications for understanding how
irrelevant stimuli affect mechanisms of eye movement control during reading.
Eye Movements During Silently Reading
Previous descriptions of reading in normal situations provide a basis for examining how
reading can be affected by music. Successful reading comprehension proceeds through a series
of stages of information processing, of which an early stage is the recognition of individual
words from printed text (Engbert, Nuthmann, Richter, & Kliegl, 2005; Reichle, Pollatsek, Fisher,
& Rayner, 1998; Reichle, Warren, & McConnell, 2009). A robust word frequency effect is often
observed in silent reading comprehension, such that low frequency words usually receive longer
looking time than do high frequency words (Rayner, 1998). Studies have suggested that word
frequency has a direct and early effect on oculomotor control during first-pass reading (Rayner,
Sereno, & Raney, 1996; Reingold, Reichle, Glaholt, & Sheridan, 2012). Further, a number of
studies have found an interaction between word length and word frequency for skilled readers,
such that the word length effect is stronger for low frequency words than for high frequency
words (e.g., Kliegl, Nuthmann, & Engbert, 2006; Pollatsek, Juhasz, Reichle, Machacek, &
Rayner, 2008; Schad, Nuthmann, & Engbert, 2012). These results can be explained by dual-route
models for visual word recognition (e.g., Coltheart, Rastle, Perry, Langdon, & Ziegler, 2001;
Paap & Noel, 1991). Dual-route theories propose two independent routes for individual word
processing: the lexical orthographic route and the sublexical phonological route. The lexical
route searches for a direct match between the printed word’s orthography and units in the mental
orthographic lexicon. Successful matching facilitates subsequent semantic and phonological
processing. Unsuccessful matching requires the word to be decoded in a serial and effortful
sublexical route based on grapheme-to-phoneme correspondence (Coltheart et al., 2001).
Although variants exist, dual-route theories agree that the sublexical route is too slow to be
relevant to the processing of all but very low frequency words. Therefore, dual-route theories
would predict an interaction between word length and word frequency, such that the recognition
of words with lower frequency should be more sensitive to word length.
After word recognition, a crucial task for reading comprehension is to integrate
individual words into higher-level representations of the text. These higher-level comprehension
processes can occur during first-pass reading. For example, gaze durations are longer on a word
when it ends a sentence/clause than when it does not (Just & Carpenter, 1980; Rayner, Kambe, &
Duffy, 2000). Just & Carpenter (1980) suggested that this “wrap-up” effect occurs because
sentence/clause end words define processing units, and skilled readers pause to integrate
meaning of the sentence/clause before moving on. After first-pass reading, readers also
occasionally go back to previously-read words. It has been generally observed that more re-
reading is associated with more in-depth processing of the text (Rayner, Chace, Slattery, &
Ashby, 2006; Schotter, Bicknell, Howard, Levy, & Rayner, 2014; Weiss, Kretzschmar,
Schlesewsky, Bornkessel-Schlesewsky, & Staub, 2017; White, Warrington, McGowan, &
Paterson, 2015). Therefore, successful re-reading may indicate the reader’s capability to monitor
online comprehension process (comprehension monitoring, Palincsar & Brown, 1984) and
interrupt progressive reading when difficulties arise. Regressive fixations were sometimes
discarded, under the assumption that they do not reflect optimal reading (Reichle et al., 2009).
For the current study, however, looking at the re-reading process may indicate to what extent
listening to music increases processing difficulties during reading.
Possible Eye Movement Patterns During Reading with Music
Although our focus is on auditory distractions, reading comprehension may be affected
by other factors as well, including top-down modulation of reading strategies (e.g., target word
searching, Rayner & Raney, 1996; skimming, Just, Carpenter, & Woolley, 1982; word
verification, Radach, Huestegge, & Reilly, 2008; topic-scanning, White et al., 2015), or a
disruption from self-generated thoughts (mindless reading; Reichle, Reineberg, & Schooler,
2010). To what extent does reading with music resemble these situations? During topic-scanning,
visual search, etc., fully comprehending the text is not necessary. During mindless reading,
although participants are asked to read for comprehension, task-irrelevant thoughts
spontaneously generated in the mind can derail reading. Therefore, certain cognitive processes
crucial for reading comprehension might be missing in these situations. Careless reading was
generally associated with a reduced word frequency effect and less re-reading, which may signal
shallow word processing and decreased comprehension efforts (Just et al., 1982; Radach et al.,
2008; Rayner & Raney, 1996; Reichle et al., 2010; Uzzaman & Joordens, 2011; White et al.,
Reading with music, however, may be different from the above scenarios. Readers are
asked to comprehend the text despite being exposed to music. In this case, human attention
attempts to maintain the current task goal and suppresses irrelevant stimuli. If the reader
understands that music is incompatible with reading comprehension, the continued presence of
this exogenous distractor may alert the reader that reading comprehension is somehow affected,
which would enable the reader to make necessary oculomotor adjustments to repair reading. This
is not the case for mindless reading, however, as the reader’s “meta-awareness” often fails to
monitor their attentional state (Schooler, Reichle, & Halpern, 2004), leading to a decoupling of
attention from perceptual information (Smallwood, 2011). Therefore, eye movement patterns
during reading with music might be different from those in the aforementioned scenarios. We
expect that crucial processes in word recognition (as measured by the word frequency effect) are
in general not affected when music is playing, but may be prone to disruption when the word is
rare, or when the music is particularly distracting. We also expect that readers would make
compensatory attempts such as re-reading to recover from processing errors. We elaborate these
points below.
As readers go through the text, they must recognize words varying in lexical complexity.
Therefore, the interference of music might depend on the processing ease of individual words,
which can be indexed by word frequency. For skilled readers, recognition of high-frequency
words is automatic and might not be strongly affected. On the other hand, music would interfere
with processing low-frequency words. According to dual-route theories, the sublexical route
requires more controlled processing than does the lexical route, as readers must correctly parse
the entire word into subunits from various potential combinations in order to construct the word
based on grapheme-to-phoneme correspondence (Coltheart et al., 2001; Paap & Noel, 1991).
Therefore, the existence of an irrelevant auditory signal may limit attentional resources that can
be allocated to the sublexical route, which would in turn disrupt processing of low frequency
words. On the other hand, music should have minimal effects on words that can be processed
through the highly automated lexical route.
Beyond recognition of individual words, the integrative processing of the text can be a
highly constructive and demanding process and therefore may be susceptible to interference. The
exposure to irrelevant auditory stimulus might create more processing difficulties that require the
reader to revisit the text. To the extent that readers are actively trying to make sense of the text,
music exposure should lead to more regressions to repair the interference than is the case for
visual search, topic-scanning, mindless reading, etc.
Reading with music should have a different effect on eye movements than does mindless
reading. In the case of reading with music, the reader’s attention is focused on perceptual stimuli
rather than internal thoughts, so interference on reading should be more closely coupled with the
dynamics of the musical stimulus. If so, we should be able to identify critical points in a piece of
music that are more likely to affect reading comprehension. Two candidates may be the onset of
a new song and the onset of the chorus. Previous studies have shown that attention to the focal
task can be involuntarily captured by novel and abrupt auditory onsets (Berti, 2013; Dalton &
Lavie, 2004). The onset of a new song produces a significant transition in the coherence of the
music stream, since the dynamics of the new song are typically very different from that of the
previous one (e.g., different genre, different voice, different pitch, etc.). The onset of the chorus,
on the other hand, represents a significant transition within one piece of music. The chorus often
contrasts sharply to the verse with its distinctive acoustic features, which allows itself to be
identified from the entire music stream (Foote & Cooper, 2003; Goto, 2006). Therefore, the
onset of a new song and the onset of the chorus might be two highly salient acoustic events that
interfere with reading and produce anomalies in eye movements.
The Current Research
The current research investigated how exposure to music affects college students’ eye
movement patterns when they read academic-style passages. Our strategy is simple: comparing a
reading episode hypothesized to be interfered by music to a less-interfered baseline, in order to
search for differences in eye movement measures between these conditions that show the effects
of interference. In Study 1, reading with self-selected music was compared to reading in silence.
In Study 2, we controlled the music participants heard, which enabled us to identify points in the
music that are more likely to produce distracted reading episodes. If effects of music on reading
depend on dynamics of the auditory signal, eye movements shortly after critical points in the
music should be disrupted.
We also sought to explore the role of extreme values in the distraction effects of music by
looking at ex-Gaussian distributions. The ex-Gaussian distribution is a convolution of a normal
distribution (represented by µ, the mean, and σ, the standard deviation) and an exponential
distribution (represented by τ, the exponential parameter that captures the skewness). Research
has shown that the different components in the ex-Gaussian distribution can indicate unique
information about cognitive processes (McVay & Kane, 2012; Staub, White, Drieghe, Hollway,
& Rayner, 2010; Unsworth, Redick, Lakey, & Young, 2010). If music interference causes a
global change in the reading process, a shift in the mean component should be expected. This
would imply that there is a psychological process independent of those involved in reading
comprehension that affects looking time on most words. However, if the reading process is only
occasionally affected by music, a shift in the exponential component should be expected, which
would imply that the effect of music is limited to a subset of words.
The reading material used in this research was chosen to reflect what college students
normally would read during college learning. Readers would occasionally encounter words that
are very low in frequency (e.g., “epiphyte”), which could increase the likelihood of sublexical
processing. Our analyses looked at the entire passages rather than focusing on specific target
words. Given the intercorrelated nature of word length and word frequency, previous studies
have often employed an orthogonal design using target words. Although evaluating the effect of
word frequency independent from word length (as well as other text features) is undoubtedly
important, it might be also helpful to know if these effects can be generalized. This is especially
important to the current study, because we are interested in dynamic relations between reading
and music over time. One can use statistical control methods such as multiple regression to
assess the effects of various predictors on fixation durations. In the current study, we adopted a
method proposed by Lorch & Meyers (1990; method 3) for multiple regressions with repeated
measures that takes both item variability and subject variability into consideration (also see
Juhasz & Rayner, 2003; Kliegl, Grabner, Rolfs, & Engbert, 2004; Kliegl et al., 2006).
Specifically, a multiple regression can be built for each participant in each condition. The
averaged regression coefficients can be tested against zero and compared between conditions. A
significant difference of a predictor between groups translates to an interaction between group
and this predictor, in ANOVA terms. One advantage of this approach is that the outcome models
take into account the entire set of fixational data and can be used to generate new predictions.
One disadvantage is that variance shared between predictors (e.g., word length and word
frequency) cannot be used for statistical tests of effects, leading to more conservative tests
compared to experiments with orthogonal designs (Kliegl et al., 2004).
Study 1
Sixty-three undergraduate students (average age = 18.51 years; thirty-six females) from a
Midwestern university participated in this study for course credit. All participants were native
English speakers and had normal or corrected-normal eyesight. None of them were familiar with
the text used in this study. The research protocol (Study 1 & Study 2) has been approved by the
institutional review board committee at the authors’ institution.
Apparatus and Stimuli
Before the experiment, participants were asked to bring their own music playlist (stored
on their own smartphone) to the experiment. During the experiment, participants played music of
their own choosing that contained English-language lyrics. Participants wore headphones and
played the music at a volume that felt comfortable to them.
Binocular eye movements were recorded with a Tobii T60 eye tracker on an embedded
17-inch display screen, at a viewing distance of approximately 60 cm, with each letter extending
horizontally across approximately .61° of visual angle. We employed the algorithm from
OGAMA 5.0 (Vosskühler, Nordmeier, Kuchinke, & Jacobs, 2008), a dispersion type algorithm
with a moving window, for event detection.
Participants were asked to read for comprehension and ignore the music. Reading stimuli
were adapted from six SAT reading comprehension practice passages. We divided the six
passages into two passage sets (A and B), and let participants read one set with music, and the
other set in silence. Thus, the music condition and the silence condition had the same number of
randomly assigned participants reading the same material, allowing us to directly contrast
reading eye movements. A and B each consisted of three passages with roughly the same total
words (1470 words for A, 1487 words for B); a total of 16 comprehension questions were
developed for each set. Set A and set B were also comparable in several lexical and linguistic
features: The average of word length (A: M = 5.04, SD = 2.66; B: M = 5.06, SD = 2.64), the
average of log (base 10) word frequency (A: M = 5.25, SD = 1.52; B: 5.17 M =, SD = 1.57) and
sentence length (A: 24 words; B: 25 words on average).
Design and Procedure
Half of the participants read set A with music playing and read set B in silence. The other
half read set A in silence and read set B with music. The reading sequence was also
counterbalanced within each group, with half of the group reading A first and the other half
reading B first. Participants were asked to read passages for comprehension and to ignore the
audio. After reading each passage, participants answered corresponding comprehension
questions. Music was played only during reading. This was accomplished by a screen prompt
asking participants to pause/resume the music using the headphone’s pause key after/before each
passage. After reading each passage set, participants also rated to what extent they felt tired from
1 (not at all) to 7 (extremely). After finishing the first passage set, participants took a short break
before moving on to the next set. Reading time was not limited.
Eye movement analyses
Fixations greater than 1200 ms and less than 80 ms were discarded. A manual drift
correction was conducted. Moreover, we noticed that participants occasionally tended to first
look over the entire page before carefully reading. These “scanning” fixations were identified
and deleted. Finally, for each participant, a page was discarded if the total count of first-pass
fixations on that page were less than 10% of the total word count of that page. Altogether, the
screening procedure resulted in a 4.45% data loss.
We first examined mean differences of various global eye movement measures between
music and silence conditions using subject-wise (F1) and item-wise (F2) comparisons. Next, ex-
Gaussian analysis of fixations were conducted using the quantile maximum likelihood estimation
algorithm (QMPE; Cousineau, Brown, & Heathcote, 2004; Heathcote, Brown, & Mewhort,
2002). The algorithm uses maximum likelihood estimation to determine the distributional
parameters that converge best to the empirical quantile distribution. This fitting process
generates three parameters, µ (the mean component), σ (the standard deviation component), and τ
(the exponential component). We conducted an ex-Gaussian estimation for data of each
participant in each condition. Output parameters were compared between conditions. Finally, we
conducted a multiple regression analysis to examine the effects of lexical and linguistic features
on fixation durations. Gaze durations (sum of all first-pass reading fixations for each word) and
total viewing times (sum of all fixations for each word) were selected as predicted variables to
reflect early and complete effects, respectively. Individualized regression models were estimated
in the music and the silence condition with the same set of predictors (Lorch & Meyers, 1990).
The means of these unstandardized coefficients were first tested against zero using one-sample t-
tests and were then compared between conditions using repeated-measure ANOVAs.
Behavioral data. The music condition produced significantly longer reading time (M =
8.37 min, SD = 3.04), compared to the silence condition (M = 7.65 min, SD = 2.40), F (1, 61) =
11.85, p <.01, η² = .16. However, reading performance (percentage of correct answers) in music
(M = 58.83%, SD = .19) was only marginally different from that in silence (M = 62.70%, SD
= .16), F (1, 61) = 3.06, p = .09, η² = .05. Moreover, tiredness rating in the music condition (M =
4.25, SD = 1.79) did not differ from that in the silence (M = 4.44, SD = 1.72), p = .34. Finally,
Participants did not feel more tired after completing the second passage group (M = 4.37, SD =
1.66) than after completing the first passage group (M = 4.33, SD = 1.85), p = .86.
Eye tracking data. We first tested the mean differences for various global measures
using repeated-measure ANOVAs. Subject-wise (F1) and item-wise (F2) results are shown in
Table 1. First-pass reading time was greater in the music condition, but the difference was not
significant. However, music condition produced significantly greater re-reading time compared
to the silence condition. The regression rate was also significantly greater in the music condition.
The skipping rates were higher than is typical for studies that look at target words in sentence
reading (e.g., Rayner, Slattery, Drieghe, & Liversedge, 2011), but are comparable to results from
a recently-developed corpus based on natural passages (about 44% in the Provo corpus, Luke &
Christianson, 2017). We note that it might be the large number of functional words in natural
passages that resulted in such high skipping rates. The materials used in the current study has
44% of functional words, with a skipping rate of over 70% (averaging both conditions). For
content words, however, the overall skipping rate was only 29%.
Next, we conducted ex-Gaussian analyses to examine the distributions of gaze durations
and total viewing times. Data from each participant in each condition was fed into the QMPE
algorithm (Cousineau, Brown, & Heathcote, 2004; Heathcote, Brown, & Mewhort, 2002) to
estimate µ, σ, and τ, which were then tested for mean differences using repeated-measure
ANOVAs. For gaze durations, no significant difference was found in the overall mean (Table 1)
or the estimated parameters (Table 2). For total viewing times, a significant difference was found
in the overall mean (Table 1), and only in τ among the three distributional parameters (Table 2).
These results indicate that the overall mean difference in total viewing times can be solely
explained by increased looking times on a subset of words.
Finally, we investigated effects of lexical and linguistic features in the music and the
silence condition. (1) We regressed gaze durations and total viewing times on the following
predictors for each participant in each condition: word length (centered on mean to reduce
multicollinearity), logarithm (base 10) of word frequency
, length by frequency interaction,
whether a word ends a sentence/clause (1 - end, 0 - not end), and whether a word is novel (1 -
duplicate, 0 - novel; reverse-coded to reflect first-time appearance). (2) We used one-sample t-
tests to examine whether the mean of unstandardized coefficients differed significantly from
zero, and (3) we used repeated-measure ANOVAs to examine whether they significantly differed
between conditions. For both conditions, all coefficients examined were significantly different
from zero, t (62) = 2.81 (or greater), p < .01. Therefore, even in the music condition lexical and
linguistic predictors made significant and independent contributions.
Results were then compared across conditions (see Table 3). The intercept did not differ
significantly for gaze durations, but the difference was significant for total viewing times.
Therefore, the model predicted that in the music condition low frequency words (at a log
frequency of zero) would receive more re-reading after first-pass reading. Importantly, results
showed that in the music condition gaze durations would be less predicted by word length when
word frequency is low. This result is supported by a significant difference of the interaction term,
indicating that the effect of word length was less modulated by word frequency compared to the
silence condition. Interestingly, these differences seemed to be reduced by continued processing
of the text, as shown by results of total viewing times.
The “wrap-up” effect also differed significantly across conditions for both gaze durations
and total viewing times. Similar to Reichle et al. (2010), we also observed seemingly reversed
“wrap-up” effects, as suggested by the negative coefficients. Note that these results might
indicate a suppression effect by the regression model (ending words were read faster than
Corpus of Contemporary American English (COCA; Davies, 2009), one of the largest currently
available corpora of American English. Some low frequency words in our reading material are
included in the COCA but not in other corpora, such as CELEX.
predicted from the addictive effects of other predictors) rather than a truly reversed “wrap-up”
effect (e.g., for gaze durations, Music: Mend = 271.04, Mno-end = 268.65; Silence: Mend = 275.08,
Mno-end = 267.25). Therefore, it is perhaps more proper to interpret these results in relative terms
(an interaction between condition and the predictor, in ANOVA terms), such that the “wrap-up”
effect was significantly weaker in the music condition relative to the silence condition.
Results from the regression analysis are visualized in Figure 1. In Figure 1 (left),
individual multiple regression models in the music and the silence condition were used to predict
gaze durations at different values of the predictors. The following values were used: Word length
(raw): Short (4) and Long (10); Log frequency: HF (high frequency, 6) and LF (low frequency,
0); Sentence/Clause end = 0; Novel = 0. It can be observed that the word length effect for low
frequency words was smaller in the music condition compared to the silence condition. This
pattern was further supported by the actual data, as shown in Figure 1 (right). Actual gaze
durations of words below 10% quantile of log word frequency were selected and plotted against
word length (<=4 ~13+). These plots visually confirmed results from the multiple regression
analysis: the word length effect for low frequency words was smaller in the music condition
compared to the silence condition.
Study 1 examined how listening to music affects the reading process by contrasting
reading with music and reading in silence. Compared to the silence condition, participants in the
music condition spent significantly more time reading the materials but still performed slightly
worse in the comprehension test. The increased reading time was reflected by significant
differences in re-reading measures. The fact that the comprehension score was slightly lower and
that more re-reading occurred indicate that these re-reading behaviors primarily served to
compensate for music’s interference.
Ex-Gaussian analysis showed that a late effect on the distribution of total viewing times
caused a global mean difference, suggesting that a subset of words received extensive re-reading
presumably due to increased processing difficulties (Staub & Benatar, 2013). Results of gaze
durations did not show any significant difference in the distributional parameters, which is
perhaps not too surprising given the finding that cases in which first-pass reading was affected
could be rare (as discussed below).
We also examined the effects of lexical and linguistic features on gaze durations and total
viewing times. Results in the silence condition replicated effects typically found in the literature
(e.g., Juhasz & Rayner, 2003; Just & Carpenter, 1980; Kliegl et al., 2004; Rayner, 1998; Rayner
et al., 2000). In the music condition, first-pass reading in general demonstrated a very similar
pattern to that in silently reading, suggesting that some oculomotor control mechanisms related
to word recognition in general remained functional despite of music exposure. This might reflect
the fact that, for skilled readers, recognition of familiar words is highly automatic. On the other
hand, results showed that the processing of very low frequency words was disrupted - when
sublexical processing was relied on, this less automated process was somehow disrupted by
irrelevant auditory signals. Interestingly, results for total viewing times suggest that these low
frequency words tended to receive more re-reading, which appear to repair the disrupted first-
pass reading. Therefore, at least one reason for increased re-reading might be to reprocess rare
words, which could contribute to the increased skewness in the distribution of total viewing time.
In the multiple regression analysis, we also found that the normal “wrap-up” effect was
significantly smaller in the music condition. Because this semantic integration process demands
attentional resources (Miller & Stine-Morrow, 1998; Payne & StineMorrow, 2012), it might be
possible that readers adopted a smaller processing unit than a whole sentence/clause to
compensate for music’s interference. Another possibility is that increased re-reading in the
middle of sentence made ending words less important. Finally, the study also found that the
effect of novel words did not differ between music and silence - words tend to receive longer
looking times on initial appearance compared to subsequent appearances.
Study 1 adopted a global perspective in which eye movements in the entire music
condition were compared with those in the silence condition. However, as mentioned before,
music may not have monolithic effects throughout the reading process and it is possible that
some reading episodes heavily disrupted by music were masked by other episodes in which little
interference occurs. If this is the case, then a global contrast between music and silence condition
does not help to pinpoint exactly where the strongest interference occurs. Therefore, we
conducted another study to examine if global findings in Study 1 can be connected to specific
musical events that might produce immediate and strong interference to reading, namely the
onset of a new song and the onset of the chorus.
Study 2
Fifty-two undergraduate students (average age = 19.19 years; seventeen females) from a
Midwestern university participated in this study. All participants were native English speakers,
have normal or corrected-normal eyesight, and none of them were familiar with the text used in
this study.
Stimuli and Procedure
Participant were asked to read eight passages for comprehension in this study. Besides
the six passages used in Study 1, participants also read two additional passages adopted from
SAT reading comprehension practice materials. These two passages have 598 and 601 words,
4.36 and 4.57 average word length, 5.57 and 5.58 log (base 10) word frequency, and 15.74 and
19.39 words per sentence, respectively. Because performance score is not the primary focus of
this study, after each passage participants answered a reduced set of comprehension questions
(two questions for each passage). Participants read all passages with music playing through
headphones. To monitor and synchronize the progress of audio stream and eye movements, we
asked participants to listen to experimenter-selected music that was played by the computer. We
selected 18 popular music pieces from Billboard (see Appendix for a list of songs). These music
pieces all have a “verse-chorus” structure and were edited so that each piece only contained the
first verse and the first chorus (i.e., the music changed to the next song after the first chorus). The
average length of a music piece was 72.15 seconds (range from 50.17 seconds to 104.26
seconds). For each music piece, the onset time of chorus was marked by two research assistants
who were familiar with the music selected. The average onset time of chorus was at 42.44
seconds (ranging from 25.73 seconds to 65.01 seconds). The average length of a chorus was
30.32 seconds (ranging from 20.11 seconds to 55.35 seconds). Each piece of music was followed
by either a one or two-second silence interval to mimic real song transitions. The music order
and the passage order were randomized for each participant. Participants were asked to read for
comprehension and to ignore the audio. Reading time was not limited. Participants’ eye
movements were recorded in an identical situation to that in Study 1. The progress of reading
and the progress of the music were monitored during the experiment and were then synchronized
after the experiment using self-developed Python scripts.
Eye movement analyses
Data screening procedure was the same as in Study 1 (5.24% of data loss). In order to
reflect the time course of distraction effects around onset, timestamp of fixations was
synchronized with the music stream and categorized into two-second time bins ranging from two
seconds before each onset to ten seconds after each onset. Fixations were allocated to bins based
on their starting times (i.e., “bridging” fixations were put into the previous bin). Individualized
multiple regression was computed for each bin Because both the order of passages and the order
of music were randomized, and readers read at their own pace, the corresponding bin for
different readers might represent different parts of the text. This ensures that results were not due
to a special combination of the music sequence and the passage sequence. Similar approaches
were adopted by some previous studies on mindless reading (e.g., Reichle et al., 2010).
Specifically, participants were randomly probed to report mind-wandering during reading, and
eye movements prior to the probes were collectively analyzed using arbitrary time windows. The
underlying assumption is that episodes of mindless reading may converge to display some
common characteristics that distinguish them from normal reading as well as random variance.
In the current study, a small bin size was chosen for depicting a refined time course of eye
movements. This might create noise in our data. However, if the effect of musical feature is
immediate and strong, episodes from different locations should overcome potential noise and
collectively demonstrate similar characteristics.
A related issue is to how to statistically examine the effect of onsets. Note that because
the bin size was arbitrarily chosen, results do not imply precisely how long the effect would
“persist”. However, if interference is immediate and strong, its effects must be at least reflected
through a short period immediately after the onset, namely the 0~2 sec bin. Specifically, value in
the 0~2 sec bin (1) must be greater than the value in the -2~0 sec bin (so that there is a local
“fluctuation”), and (2) must be greater than an average value representing music’s effect on the
global reading session (so that the effect “stands out” from the baseline as global “outliers”). To
avoid over-interpreting our results, we only examined data in the first bin (e.g., 0~2 sec) for each
measure. Data in the 0~2 sec bin were compared to those in the -2~0 sec bin, as well as those
calculated by using data of the entire reading session. Data in the subsequent bins are shown, but
not tested for significance.
Although the bin approach allows us to relate musical events and reading parameters,
there are some limitations to this approach. Because eye movement measures are calculated
separately for each bin, some eye movement measures cannot be computed accurately. For
example, total viewing times may be biased because participants may still fixate back after
passing a bin’s right boundary. Also, it is difficult to unambiguously assign skipped words to
bins. Finally, given the constant size of bins, there is an inverse relationship between the duration
of fixations and the number of fixations within a bin. Therefore, we restricted our analyses to
measures minimally affected by the segmentation. We report mean of gaze durations (by
merging first-pass fixations on the same word), mean of individual re-reading fixations,
regression rate, and percentage of first-fixations out of all fixations (a measure of new words
being read within a specific time range). We also examined effects of lexical and linguistic
variables on gaze durations for each bin, using the same set of predictors as in Study 1. For every
measure examined, a baseline value was calculated from the entire data. The baseline was used
to examine if values in the 0~2 sec bin can “stand out” from the global effect of music as a
global “outlier”. This comparison provides a very conservative test as to whether anomalies in
eye movements can be exacerbated when the music reaches certain onset.
The time course of several global eye movement measures is shown in Figure 2. For
convenience, we denoted all local comparisons (with the -2~0 sec bin) as tlocal, and all
comparisons with the global baseline as tglobal. For the percentage of first-fixations (Figure 2,
panel a), the 0~2 sec bin after new song onset was significant for both tlocal (51) = 6.53, p < .001,
d = -.90, and tglobal (51) = 4.15, p < .001, d = -.58; However, the chorus onset did not have
significant differences during the same interval, p > .05. For gaze durations (Figure 2, panel b),
the 0~2 sec bin after new song onset differed significantly, tlocal (51) = 4.58, p < .001, d = .63,
and tglobal (51) = 6.32, p < .001, d = .88; However, the chorus onset did not have significant
differences during the same interval, p > .05. For the regression rate (Figure 2, panel c),
significant differences were found for the new song onset, tlocal (51) = 4.16, p < .001, d = .58, and
tglobal (51) = 2.71, p < .01, d = .38. For the chorus onset, both tlocal and tglobal did not reach
significance, p > .05. Finally, for the mean of re-reading fixations (Figure 2, panel d), the new
song onset again yielded significant differences, tlocal (51) = 3.97, p < .001, d = .55, and tglobal
(51) = 4.25, p < .001, d = .59. For the chorus onset, only tglobal reached significance, tglobal (51) =
2.42, p < .05, d = .33. Therefore, we concluded that these global eye movement measures were
seriously affected by the new song onset but not the chorus onset.
The significant increase in gaze durations after new song onset were further examined by
ex-Gaussian analysis. Results (Table 4) showed that the overall difference can be solely
explained by an increased weight on the distribution’s tail, indicating that new song onset only
increased the skewness of the distribution.
Finally, we examined how new song onset influenced lexical and linguistic effects on
gaze durations. Individualized multiple regressions (Lorch & Meyers, 1990) were conducted for
each bin as well as for the entire reading session, with the same predictor set as in Study 1. The
time course of these effects (unstandardized coefficients averaged across participants) are shown
in Figure 3. For the intercept (panel a), tlocal was not significant, p > .05, and tglobal was significant
at .05 level, tglobal (51) = 2.04, p = .046, d = .28. For the word length effect (panel b), both local
and global comparisons showed a significant decrease, tlocal (51) = 3.70, p < .001, d = -.51, tglobal
(51) = 2.23, p < .05, d = -.31. Importantly, during 0~2 sec the word length effect itself was not
significantly different from zero, p > .05. Consistent with these results, the length by frequency
interaction effect (panel d) was also significantly decreased, tlocal (51) = 4.13, p < .001, d = .57,
tglobal (51) = 2.42, p < .05, d = .34. Again, during 0~2 sec the interaction term did not differ from
zero, p > .05. For the word frequency effect (panel c), both tlocal and tglobal did not reach
significance, p > .05. Remarkably, the word frequency effect itself was still significantly
different from zero, t (51) = 2.62, p < .05. Panel e and f showed results of the “wrap-up” effect
and the novel word effect, respectively. For both effects, global and local comparisons did not
yield significant differences.
In Study 2, we further investigated distraction effects of music by examining the time
course of various eye movement measures at locations that were hypothesized to be highly
distracting, namely the onset of a new song and the onset of the chorus. We organized data into
time bins and examined whether the two types of onset can produce immediate and strong
interference on various eye movement measures. Because data were compared to the music
condition “itself”, it provides a conservative test as to if some parts of the music can produce the
strongest interference effects.
During a short period after new song onset, participants produced more re-reading, and
thus less progressive reading. These results suggest that the onset of new song had an immediate
and strong effect on producing processing difficulties that require the reader to go back. Another
important finding is that the effect of word frequency was not severely affected by the new song
onset, whereas the effect of word length and length*frequency interaction became significantly
smaller (and not significantly different from zero) immediately after onset. Ex-Gaussian results
on gaze durations did not become significant in Study 1, possibly because these differences were
revealed only when strong distraction episodes were precisely located. In general, results showed
that erratic eye movements occurred immediately when a new song started, during which eye
movement patterns were qualitatively consistent with findings of Study 1.
The results did not demonstrate a similar pattern for the chorus onset. Although the
chorus has distinct acoustical features from the verse, it may be also highly familiar and
predictable. Therefore, the onset of chorus might be a weaker and less consistent distracting
point compared to the onset of a new song. However, it is improper to conclude that the chorus
was not distracting at all, because this would have required comparison with reading in silence.
General Discussion
The practical implications of this study are relatively straightforward: listening to music
makes reading comprehension less efficient (if not worse), especially when the material is
difficult, and the music contains highly contrasting dynamics. But looking at the specific ways
that music disrupts reading has theoretical implications for understanding both reading and the
performance of complex tasks in disrupting settings.
How music affects reading
Word recognition. Reading comprehension with music did not make first-pass reading more
careless, which is different from cases like visual search (Rayner & Raney, 1996), skimming
(Just et al., 1982), word verification (Radach et al., 2008), and mindless reading (Reichle et al.,
2010). Reilly & Radach (2006) suggest that top-down modulation of reading strategies can be
simulated by changing a global activity threshold that affects the timing of saccade triggering.
Whereas this threshold might be lowered in the above cases, the current results suggest that
music exposure does not lead to this kind of effect. Note that the current research adopted a
multiple regression approach, using all fixated words to estimate effects. The fact that a global
word frequency effect exists beyond word length and other predictors is consistent with the
direct lexical account of eye movements during reading, which claims that word frequency exerts
immediate control over most fixations (e.g., Rayner et al., 1996; Reichle et al., 2003; Reingold et
al., 2012). Moreover, this mechanism seems to function similarly for reading comprehension
with music and in silence. This may reflect the fact that word recognition for skilled readers is
highly automated and therefore less susceptible to external interference.
The current study also found a robust interaction between length and frequency, which is
consistent with what dual-route theories would predict (e.g., Coltheart et al., 2001; Paap & Noel,
1991). One reason for this effect might relate to the materials used - the current material contains
a relatively high number of very rare words. The finding that a pattern similar to that observed in
a single word naming task was found in eye movements during a reading comprehension task is
consistent with the notion that a similar division between orthography- and phonology-based
mechanisms also exists in print to meaning (for a discussion, see Harm & Seidenberg, 2004).
This dual-route pattern was impaired by music, especially during high saliency periods. Previous
studies have noted several cases in which the sublexical route can be more severely impaired
compared to the lexical route (Bernstein & Carr, 1996; Herdman & Beckett, 1996; Paap & Noel,
1991). For example, it has been shown that the sublexical route can be selectively prolonged by
increasing memory load, which could lead to a release-from-competition effect that speeds up
the naming of low frequency exception words (Paap & Noel, 1991). The current results appear to
fit this account, as a similar observation with word length was found in Study 1. Therefore, to the
extent that a similar dual-route architecture exists in reading, distracting segments of the music
could capture the reader’s attention, selectively affecting the sublexical route. The lexical route,
on the other hand, would be minimally affected due to its high automaticity. In this sense, it may
be important to look at how individual differences in reading and executive functions can
modulate the current findings. The interference of music might be more severe for readers who
rely more on sublexical processing (e.g., children; Tiffin-Richards & Schroeder, 2015) and for
readers who have lower executive function capabilities.
Higher-level processing. Readers exposed to music produced more re-reading. This finding
contributes to a growing set of factors that affects re-reading in reading comprehension,
including conceptual difficulty (Rayner et al., 2006), comprehension demand (Weiss et al.,
2017), language proficiency (Reichle et al., 2013), font difficulty (Rayner, Reichle, Stroud,
Williams, & Pollatsek, 2006), etc. Due to extensive re-reading, we did not find readers
comprehension scores to be significantly impaired by music. Dixon & Li (2013) proposed that
skilled readers infer the amount of attentional resources allocated to reading comprehension by
examining the richness of situational model in their working memory. An impoverished
situational model could lead to the inference that attention is distracted by music, triggering re-
reading behaviors. The current research further demonstrated that the decision to re-read the text
can be initiated very quickly (even in a somewhat automatic manner) upon disruptions, given the
finding that re-reading was immediately initiated in response to the new song onset. Therefore,
deciding appropriately when to move eyes regressively might indicate successful comprehension
overall and should become a crucial component in models of oculomotor control during reading.
Previous models of oculomotor control of reading provide different explanations for
rereading, although both are consistent with our results. The SWIFT model (Engbert et al., 2005)
assumes that the majority of regressions are triggered by incomplete recognition of words. The
current research showed that the reduced length effect on rare words during first-pass reading is
restored for re-reading. This suggests that one reason for re-reading text might be to reprocess
rare words so that the total looking time on these words becomes length-sensitive. On the other
hand, the E-Z reader model (Reichle et al., 2009) assumes that the majority of regressions are
due to difficulties in post-lexical processing (I), rather than word recognition. Regressive
movements of attention (and/or saccades) are initiated when the integration of word n (1) fails to
complete before the lexical processing of word n+1 completes, or (2) is terminated due to an
detection of an occasional integration failure (i.e., rapid integration failure). In this sense, word
recognition problems during first-pass reading are likely to be detected during the post-lexical
integration (e.g., integrating words into sentences), triggering regressions. Moreover, because
eye movements are only affected when the relevant component fails, its effect is only linked to a
subset of fixations. Critically, E-Z reader simulations showed that effects of I can even manifest
in first-pass reading when the likelihood of rapid integration failure is high (Reichle et al., 2009).
The larger τ in gaze durations in Study 2 thus can be partly attributed to new song onset
producing severe post-lexical integration failures. In general, it seems likely that re-reading can
repair errors in both lexical processing (as suggested by the SWIFT) and post-lexical processing
(as suggested by the E-Z Reader), with the relative likelihood of each still in question. Because
reading while listening to music produces a high incidence of regressions, it will be a good
context for looking at the extent to which regressions result from word recognition failures or
difficulty in integrating the text.
Reading under External vs. Internal Distractions
Although both external and internal factors can disrupt attention to text, their eye
movement patterns seem to be different. While studies of mindless reading have documented a
breakdown of the word frequency effect and a lack of re-reading, the current research shows that
listening to music has minimal impact on the word frequency effect and produced more re-
reading. These differences seem puzzling, given recent notions suggesting that the occurrence of
both external and internal distractions share an underlying attentional control mechanism
(Forster, 2013; Forster & Lavie, 2014; McVay & Kane, 2009, 2010; Unsworth & McMillan,
2014). Here, we attempt to reconcile these results.
Recovering from a distracted state might be harder for internal distractions than for
external distractions. One pattern suggested by the current research is that cognitive processes
that are less automated are more affected by irrelevant stimuli. This pattern is consistent with
mindless reading. Schooler et al. (2004) suggested that, because word recognition for skilled
readers is highly automatic, lexical processing should be less affected than are higher-level
processes during mindless reading (e.g., Smallwood, 2011; Smallwood, McSpadden, &
Schooler, 2008). Further, Schad et al. (2012) proposed that the transition from mindful to
mindless reading should be graded, with processes involving more cognitive control lost earlier
than those that are more automatic. However, readers during mindless reading may find it
difficult to terminate this disrupted state. One reason might relate to the “temporal dissociation”
between experience and meta-awareness during mindless reading (Schooler, 2002). Smallwood
(2013) noted that working memory plays different roles before and during mind-wandering:
working memory suppresses task-unrelated stimuli during on-task periods, but once failed, it
turns inward to maintain the continuity and integrity of self-generated thoughts. During mindless
reading, readers are experientially conscious of irrelevant self-generated thoughts, meanwhile
lacking meta-awareness of the fact that they are off-task (Schooler et al., 2004). The dissociation
between meta-awareness and experience might prevent the reader from re-orienting attention
upon disruptions during mindless reading.
In the current study, however, the finding that readers’ oculomotor control was highly
sensitive to the onset of a new song suggests that readers stayed in a state of attention that
welcomes perceptual information, which allows for both reading and music interference. These
results also appear to be at odd with claims that listening to music reduces processing perceptual
information (Herbert, 2012, 2013; Schäfer & Fachner, 2015). Once distraction occurs, the reader
might be able to recover from the distracted state more successfully than is the case for mind-
wandering: during mindless reading, it is often not until quite some time later does the reader
catches herself and starts to read all over again.
The timely detection of disruptions protects
A direct test of this idea is to compare the duration of the same cognitive process being in the
distracted state by external and internal factors. This attempt, however, can be hindered by the
reading from further deteriorating to a state where even automated processes such as lexical
processing are disrupted (Schad et al., 2012), as is often observed during deep mindless reading.
Further, efforts to compensate for disruptions were permitted by the current research
design. Due to a lack of direct manipulations, previous studies observing mindless reading often
adopted the experience-sampling approach, waiting for positive self-reports of mind-wandering
and backtracking from this endpoint (e.g., Reichle et al., 2010; Uzzaman & Joordens, 2011). But
what happens after the read detects a problem might be as interesting as what happens before: if
the reader realizes that she has been going through the text mindlessly, progressive reading
should be replaced by re-reading. The current study did not attempt to interrupt reading upon
distractions and therefore might include a series of transitions between on-task and distracted
states. While certain cognitive processes (especially more controlled processes) might be
disrupted, the attempt to quickly recover from abnormalities was reflected in re-reading.
Bjork (1994) coined the term “desirable difficulty” for conditions that lead to additional
effort in learning but result in increased performance in the long run. This includes features such
as spacing and interleaving practice or providing delayed rather than immediate feedback.
Attempting to read in the presence of music did make the task more difficult as shown by
increased rereading and overall longer time, but this additional effort did not lead to increased
performance. This suggests that the difficulties produced by music should still be considered
undesirable. However, it may be premature to conclude that music plays only a negative role in
the complex ecology of factors that readers use to manage disruptions. Listening with music
playing does produce disruptions, but those are ones that skilled readers can readily manage with
fact that no reliable method is available to estimate the duration of a certain mind-wandering
period, due to its internal and spontaneous nature (Smallwood, 2013).
rereading. Perhaps the external disruptions due to music help to block internal and potentially
more serious disruptions due to mind wandering.
The current research limited auditory stimuli to music with vocal sounds and is unable to
specify which acoustic component in the music has the strongest contribution to such
interference. Tempo, pitch, volume, etc., can affect arousal level, which might influence
oculomotor control (e.g., Day, Lin, Huang, & Chuang, 2009). In Study 1, participants brought
their own music, which may reduce music’s effect on first-pass reading due to music’s high
familiarity. Moreover, semantic information contained in auditory distractors can be
involuntarily processed to influence reading, as both processes involve meaning (Jones, Miles, &
Page, 1990; Oswald, Tremblay, & Jones, 2000; although see Boyle, 1996). More importantly, the
current research is unable to distinguish effects of music from non-musical effects. Non-musical
tones with abrupt changes in tone or pitch might elicit similar/identical eye movement patterns.
Future research should separate effects from various distraction sources by carefully
manipulating characteristics of musical stimuli, and examine if current findings can be
generalized to other non-musical sounds, such as irrelevant speech or tones.
Although there are many aspects yet to explore, the current research contributes to a
growing understanding of how eye movements during reading comprehension are affected by
irrelevant factors, which may further our understanding of human regulation of attention and
awareness in different real-life task scenarios. Indeed, given the ubiquity of music and other
distractions in contexts where adults read, it may make sense to start to question whether reading
in silence in a quiet lab setting should really be called “normal” reading.
Banbury, S. P., Macken, W. J., Tremblay, S., & Jones, D. M. (2001). Auditory Distraction and
Short-Term Memory: Phenomena and Practical Implications. Human Factors: The Journal
of the Human Factors and Ergonomics Society, 43(1), 1229.
Bernstein, S., & Carr, T. (1996). Dual-Route Theories of Pronouncing Printed Words: What Can
Be Learned From Concurrent Task Performance?. J Exp Psychol Learn Mem Cogn.
Berti, S. (2013). The role of auditory transient and deviance processing in distraction of task
performance: a combined behavioral and event-related brain potential study. Frontiers in
Human Neuroscience, 7.
Bjork, R. A. (1994). Institutional impediments to effective training. Learning, Remembering,
Believing: Enhancing Human Performance, 295306.
Boyle, R. (1996). Effects of Irrelevant Sounds on Phonological Coding in Reading
Comprehension and Short term Memory. The Quarterly Journal of Experimental
Psychology. A, Human Experimental Psychology, 49(March 2013), 398416.
Calderwood, C., Ackerman, P. L., & Conklin, E. M. (2014). What else do college students “do”
while studying? An investigation of multitasking. Computers and Education, 75, 1929.
Cassidy, G., & Macdonald, R. (2009). The effects of music choice on task performance: A study
of the impact of self-selected and experimenter-selected music on driving game
performance and experience. Musicae Scientiae, 13(2), 357386.
Coltheart, M., Rastle, K., Perry, C., Langdon, R., & Ziegler, J. (2001). DRC: A dual route
cascaded model of visual word recognition and reading aloud. Psychological Review,
108(1), 204256.
Cousineau, D., Brown, S., & Heathcote, A. (2004). Fitting distributions using maximum
likelihood: methods and packages. Behavior Research Methods, 36(4), 742756.
Crawford, H., & Strapp, C. (1994). Effects of vocal and instrumental music on visuospatial and
verbal performance as moderated by studying preference and personality. Personality and
Individual Differences, 16(2), 237245.
Dalton, B. H., & Behm, D. G. (2007). Effects of noise and music on human and task
performance : A systematic review, 7, 143152.
Dalton, P., & Lavie, N. (2004). Auditory Attentional Capture: Effects of Singleton Distractor
Sounds. Journal of Experimental Psychology. Human Perception & Performance, 30(1),
Davies, M. (2009). The 385+ million word Corpus of Contemporary American English (1990
2008+). Design, architecture, and linguistic insights. International Journal of Corpus
Linguistics, 14(2), 159190. 10.1075/ijcl.14.2.02dav
Day, R. F., Lin, C. H., Huang, W. H., & Chuang, S. H. (2009). Effects of music tempo and task
difficulty on multi-attribute decision-making: An eye-tracking approach. Computers in
Human Behavior, 25(1), 130143.
Dixon, P., & Li, H. (2013). Mind wandering in text comprehension under dual-task conditions.
Frontiers in Psychology, 4(OCT).
Engbert, R., Nuthmann, A., Richter, E. M., & Kliegl, R. (2005). SWIFT: A Dynamical Model of
Saccade Generation During Reading. Psychological Review, 112(4), 777813.
Foote, J., & Cooper, M. L. (2003). Media segmentation using self-similarity decomposition. In
Proceedings of SPIE (Vol. 5021, pp. 167175).
Forster, S. (2013). Distraction and mind-wandering under load. Frontiers in Psychology,
4(MAY), 16.
Forster, S., & Lavie, N. (2014). Distracted by your mind? Individual differences in distractibility
predict mind wandering. Journal of Experimental Psychology: Learning, Memory, and
Cognition, 40(1), 251260.
Furnham, A., & Allass, K. (1999). The influence of musical distraction of varying complexity on
the cognitive performance of extroverts and introverts. European Journal of Personality,
38(January 1998), 2738. Retrieved from
Furnham, A., & Bradley, A. (1997). Music while you work: The differential distraction of
background music on the cognitive test performance of introverts and extraverts. Applied
Cognitive Psychology, 11(January), 445455. Retrieved from
Goto, M. (2006). A chorus section detection method for musical audio signals and its application
to a music listening station. IEEE Transactions on Audio, Speech and Language
Processing, 14(5), 17831794.
Harm, M. W., & Seidenberg, M. S. (2004). Computing the Meanings of Words in Reading:
Cooperative Division of Labor Between Visual and Phonological Processes. Psychological
Review, 111(3), 662720.
Heathcote, A., Brown, S., & Mewhort, D. J. K. (2002). Quantile maximum likelihood estimation
of response time distributions. Psychonomic Bulletin & Review, 9(2), 394401.
Herbert, R. (2012). Consciousness and everyday music listening: Trancing, dissociation, and
absorption. In Music and Consciousness: Philosophical, Psychological, and Cultural
Perspectives (pp. 118).
Herbert, R. (2013). An empirical study of normative dissociation in musical and non-musical
everyday life experiences. Psychology of Music, 41(3), 372394.
Herdman, C. M., & Beckett, B. L. (1996). Code-specific processes in word naming: Evidence
supporting a dual-route model of word recognition. Journal of Experimental Psychology.
Human Perception & Performance, 22, 11491165.
Jones, D. M., Miles, C., & Page, J. (1990). Disruption of proofreading by irrelevant speech:
Effects of attention, arousal or memory? Applied Cognitive Psychology.
Juhasz, B. J., & Rayner, K. (2003). Investigating the Effects of a Set of Intercorrelated Variables
on Eye Fixation Durations in Reading. Journal of Experimental Psychology: Learning,
Memory, and Cognition, 29(6), 13121318.
Just, M. A., & Carpenter, P. A. (1980). A theory of reading: From eye fixations to
comprehension. Psychological Review, 87(4), 329354.
Just, Carpenter, & Woolley. (1982). Paradigms and processes in reading comprehension. Journal
of Experimental Psychology. General, 111(2), 228238.
Kampfe, J., Sedlmeier, P., & Renkewitz, F. (2010). The impact of background music on adult
listeners: A meta-analysis. Psychology of Music, 39(4), 424448.
Kliegl, R., Grabner, E., Rolfs, M., & Engbert, R. (2004). Length, frequency, and predictability
effects of words on eye movements in reading. European Journal of Cognitive Psychology,
16(12), 262284.
Kliegl, R., Nuthmann, A., & Engbert, R. (2006). Tracking the mind during reading: The
influence of past, present, and future words on fixation durations. Journal of Experimental
Psychology: General, 135(1), 1235.
Lorch, R. F., & Myers, J. L. (1990). Regression analyses of repeated measures data in cognitive
research. Journal of Experimental Psychology: Learning, Memory, and Cognition, 16(1),
Luke, S. G., & Christianson, K. (2017). The Provo Corpus: A large eye-tracking corpus with
predictability norms. Behavior Research Methods.
McVay, J. C., & Kane, M. J. (2009). Conducting the train of thought: Working memory capacity,
goal neglect, and mind wandering in an executive-control task. Journal of Experimental
Psychology: Learning, Memory, and Cognition, 35(1), 196204.
McVay, J. C., & Kane, M. J. (2010). Does mind wandering reflect executive function or
executive failure? Comment on Smallwood and Schooler (2006) and Watkins (2008).
Psychological Bulletin, 136(2), 188197.
McVay, J. C., & Kane, M. J. (2012). Drifting from slow to “d’oh!”: Working memory capacity
and mind wandering predict extreme reaction times and executive control errors. Journal of
Experimental Psychology: Learning, Memory, and Cognition, 38(3), 525549.
Miller, L. M., & Stine-Morrow, E. a. (1998). Aging and the effects of knowledge on on-line
reading strategies. The Journals of Gerontology. Series B, Psychological Sciences and
Social Sciences, 53(4), P223-33. Retrieved from
Oswald, C. J., Tremblay, S., & Jones, D. M. (2000). Disruption of comprehension by the
meaning of irrelevant sound. Memory (Hove, England), 8(5), 345350.
Paap, K. R., & Noel, R. W. (1991). Dual-route models of print to sound: Still a good horse race.
Psychological Research, 53(1), 1324.
Palincsar, A. S., & Brown, A. L. (1984). Reciprocal Teaching of Comprehension- Fostering and
Comprehension- Monitoring Activities. Cognition and Instruction, 1(2), 117175.
Payne, B. R., & StineMorrow, E. A. L. (2012). Aging, parafoveal preview, and semantic
integration in sentence processing: Testing the cognitive workload of wrap-up. Psychology
and Aging, 27(3), 638649.
Perham, N., & Sykora, M. (2012). Disliked Music can be Better for Performance than Liked
Music. Applied Cognitive Psychology, 26(4), 550555.
Pollatsek, A., Juhasz, B. J., Reichle, E. D., Machacek, D., & Rayner, K. (2008). Immediate and
delayed effects of word frequency and word length on eye movements in reading: a
reversed delayed effect of word length. Journal of Experimental Psychology: Human
Perception and Performance, 34(3), 72650.
Radach, R., Huestegge, L., & Reilly, R. (2008). The role of global top-down factors in local eye-
movement control in reading. Psychological Research, 72(6), 675688.
Rayner, K. (1998). Eye movements in reading and information processing: 20 years of research.
Psychological Bulletin, 124(3), 372422.
Rayner, K., Chace, K. H., Slattery, T. J., & Ashby, J. (2006). Eye Movements as Reflections of
Comprehension Processes in Reading. Scientific Studies of Reading, 10(3), 241255.
Rayner, K., Kambe, G., & Duffy, S. A. (2000). The effect of clause wrap-up on eye movements
during reading. The Quarterly Journal of Experimental Psychology A: Human Experimental
Psychology, 53A(4), 10611080.
Rayner, K., & Raney, G. E. (1996). Eye movement control in reading and visual search: Effects
of word frequency. Psychonomic Bulletin & Review, 3(2), 2458.
Rayner, K., Reichle, E. D., Stroud, M. J., Williams, C. C., & Pollatsek, A. (2006). The effect of
word frequency, word predictability, and font difficulty on the eye movements of young and
older readers. Psychology and Aging, 21(3), 448465.
Rayner, K., Sereno, S. C., & Raney, G. E. (1996). Eye movement control in reading: A
comparison of two types of models. Journal of Experimental Psychology: Human
Perception and Performance, 22(5), 11881200.
Rayner, K., Slattery, T. J., Drieghe, D., & Liversedge, S. P. (2011). Eye movements and word
skipping during reading: Effects of word length and predictability. Journal of Experimental
Psychology: Human Perception and Performance, 37(2), 514528.
Reichle, E. D., Liversedge, S. P., Drieghe, D., Blythe, H. I., Joseph, H. S. S. L., White, S. J., &
Rayner, K. (2013). Using E-Z Reader to examine the concurrent development of eye-
movement control and reading skill. Developmental Review, 33(2), 110149.
Reichle, E. D., Pollatsek, a, Fisher, D. L., & Rayner, K. (1998). Toward a model of eye
movement control in reading. Psychological Review, 105(1), 125157.
Reichle, E. D., Rayner, K., & Pollatsek, A. (2003). The E-Z reader model of eye-movement
control in reading: comparisons to other models. The Behavioral and Brain Sciences, 26(4),
Reichle, E. D., Reineberg, A. E., & Schooler, J. W. (2010). Eye movements during mindless
reading. Psychological Science : A Journal of the American Psychological Society / APS,
21(9), 13001310.
Reichle, E. D., Warren, T., & McConnell, K. (2009). Using E-Z Reader to model the effects of
higher level language processing on eye movements during reading. Psychonomic Bulletin
& Review, 16(1), 121.
Reichle, E. D., Warren, T., & McConnell, K. (2009). Using E-Z Reader to model the effects of
higher level language processing on eye movements during reading. Psychonomic Bulletin
& Review, 16(1), 121.
Reilly, R. G., & Radach, R. (2006). Some empirical tests of an interactive activation model of
eye movement control in reading. Cognitive Systems Research, 7(1), 3455.
Reingold, E. M., Reichle, E. D., Glaholt, M. G., & Sheridan, H. (2012). Direct lexical control of
eye movements in reading: Evidence from a survival analysis of fixation durations.
Cognitive Psychology, 65(2), 177206.
Schad, D. J., Nuthmann, A., & Engbert, R. (2012a). Your mind wanders weakly, your mind
wanders deeply: Objective measures reveal mindless reading at different levels. Cognition,
125(2), 179194.
Schad, D. J., Nuthmann, A., & Engbert, R. (2012b). Your mind wanders weakly, your mind
wanders deeply: Objective measures reveal mindless reading at different levels. Cognition,
125(2), 179194.
Schäfer, T., & Fachner, J. (2015). Listening to music reduces eye movements. Attention,
Perception, & Psychophysics, 77(2), 551559.
Schooler, Jonathan W.; Reichle, Erik D.; Halpern, D. V. (2004). Zoning Out while Reading:
Evidence for Dissociations between Experience and Metaconsciousness. In Thinking and
seeing: Visual metacognition in adults and children (pp. 203226).
Schooler, J. W. (2002). Re-representing consciousness: Dissociations between experience and
meta-consciousness. Trends in Cognitive Sciences.
Schotter, E. R., Bicknell, K., Howard, I., Levy, R., & Rayner, K. (2014). Task effects reveal
cognitive flexibility responding to frequency and predictability: Evidence from eye
movements in reading and proofreading. Cognition, 131(1), 127.
Smallwood, J. (2011). Mind-wandering While Reading: Attentional Decoupling, Mindless
Reading and the Cascade Model of Inattention. Linguistics and Language Compass.
Smallwood, J. (2013). Distinguishing how from why the mind wanders: A processoccurrence
framework for self-generated mental activity. Psychological Bulletin, 139(3), 519535.
Smallwood, J., McSpadden, M., & Schooler, J. W. (2008). When attention matters: The curious
incident of the wandering mind. Memory & Cognition, 36(6), 11441150.
Staub, A., & Benatar, A. (2013). Individual differences in fixation duration distributions in
reading. Psychonomic Bulletin & Review, 20(6), 13041311.
Staub, A., White, S. J., Drieghe, D., Hollway, E. C., & Rayner, K. (2010). Distributional effects
of word frequency on eye fixation durations. Journal of Experimental Psychology: Human
Perception and Performance, 36(5), 12801293.
Tiffin-Richards, S. P., & Schroeder, S. (2015). Word length and frequency effects on children’s
eye movements during silent reading. Vision Research, 113(PA), 3343.
Unsworth, N., & McMillan, B. D. (2014). Similarities and differences between mind-wandering
and external distraction: A latent variable analysis of lapses of attention and their relation to
cognitive abilities. Acta Psychologica, 150, 1425.
Unsworth, N., Redick, T. S., Lakey, C. E., & Young, D. L. (2010). Lapses in sustained attention
and their relation to executive control and fluid abilities: An individual differences
investigation. Intelligence, 38(1), 111122.
Uzzaman, S., & Joordens, S. (2011). The eyes know what you are thinking: Eye movements as
an objective measure of mind wandering. Consciousness and Cognition, 20(4), 18821886.
Vosskühler, A., Nordmeier, V., Kuchinke, L., & Jacobs, A. M. (2008). OGAMA (Open Gaze
and Mouse Analyzer): open-source software designed to analyze eye and mouse movements
in slideshow study designs. Behavior Research Methods, 40(4), 11501162.
Weiss, A. F., Kretzschmar, F., Schlesewsky, M., Bornkessel-Schlesewsky, I., & Staub, A.
(2017). Comprehension demands modulate re-reading, but not first pass reading behavior.
The Quarterly Journal of Experimental Psychology, 0(0), 137.
White, S. J., Warrington, K. L., McGowan, V. A., & Paterson, K. B. (2015). Eye movements
during reading and topic scanning: effects of word frequency. Journal of Experimental
Psychology. Human Perception and Performance, 41(1), 23348.
Tables and Figures:
First-pass reading time (sec)
208.27 (6.65)
206.74 (7.01)
< .01
Gaze duration (msec)
266.33 (4.86)
264.97 (4.89)
Skipping rate (%)
47.59 (1.07)
48.01 (1.16)
Re-reading time (sec)
98.67 (6.89)
80.09 (5.56)
Total viewing time (msec)
394.14 (11.13)
368.69 (9.88)
Regression rate (%)
18.18 (.69)
17.37 (.70)
Table 2
Means (SE) of the ex-Gaussian parameters for gaze durations and total viewing times.
Test of mean
Gaze durations
130.30 (2.17)
130.22 (2.07)
32.38 (.98)
32.97 (.97)
135.82 (4.59)
134.58 (4.64)
Total viewing
126.45 (3.75)
125.04 (2.70)
29.59 (1.36)
30.49 (1.01)
262.34 (9.76)
243.54 (9.12)
Note. Data from each participant in each condition was fed into the QMPE algorithm to
estimate a set of three distributional parameters, which were then tested for mean
differences using repeated-measure ANOVAs. Participants who were not satisfactorily
converged (error code >= 33) were discarded (gaze durations 0, total viewing times - 2).
Note that for the purpose of consistency, results for total viewing times were calculated
only for a subset of words that received first-pass reading. *p < .05; **p < .01; ***p
< .001.
Table 3
Mean (SE) of unstandardized coefficients for gaze durations and total viewing times in the
music and the silence condition
Test of mean
326.82 (8.44)
318.04 (7.13)
16.84 (1.42)
21.66 (1.75)
-13.51 (1.16)
-12.66 (1.00)
-2.59 (.32)
-3.53 (.39)
-9.30 (1.73)
-11.52 (1.41)
Sentence/Clause End
-17.19 (3.79)
-9.12 (3.25)
537.90 (20.89)
494.66 (16.88)
32.63 (2.54)
35.40 (2.88)
-30.86 (2.89)
-27.91 (2.25)
-4.21 (.58)
-5.11 (.64)
-24.40 (3.99)
-23.54 (2.99)
Sentence/Clause End
-55.35 (7.61)
-38.23 (5.93)
Note. Length: centered on mean; Freq: logarithm (base 10) of word frequency; Novel: whether
it is the first appearance of a word in a passage, 1 - no, 0 - yes (reverse-coded to reflect first-
time appearance); Sentence/Clause End: whether a word is the last word of each
sentence/clause, 1 - yes, -1 - no. Multiple regression was estimated for each participant in each
condition. Note that for the purpose of consistency, results for total viewing times were
calculated only for a subset of words that received first-pass reading. Means of unstandardized
coefficients were tested by one-sample t-tests (all coefficients listed were significantly
different from zero, t (62) > 2.81, p < .01) and repeated measure ANOVAs (degree of
freedom: F (1, 61)). *p < .05; **p < .01; ***p < .001
Table 4
Ex-Gaussian parameters for the distribution of gaze durations before and after new song
Time relative to onset
Test of mean
133.21 (3.37)
135.43 (3.74)
34.67 (2.53)
35.93 (2.65)
143.68 (5.62)
164.22 (7.05)
Note. Data from each participant in each interval (before/after onset) was fed into the
QMPE algorithm to estimate a set of three distributional parameters, which were then
tested for mean differences using paired-sample t-tests. Three participants who were not
satisfactorily converged (error code >= 33) were discarded from this analysis. Degree of
freedom: t (48). *p < .05; **p < .01; ***p < .001.
Figure 1. Predicted and actual gaze durations in the music and the silence conditions. Left:
Gaze durations predicted by individual multiple regression models. The following values were
used to generate these values: Word length (raw): Short (4) and Long (10); Log frequency: HF
(high frequency, 6) and LF (low frequency, 0); Sentence/Clause end = 0; Novel = 0. Error bars
showed standard errors. Right: Actual gaze durations. Actual gaze durations of words below
10% quantile of log word frequency were plotted against word length (<=4 ~13+). Error bars
showed standard errors.
Figure 2. Time course of various global eye movement measures before and after the new
song/chorus onset. Panel a: Percentage of first-fixations (out of all fixations). Panel b: Mean
of gaze durations. Panel c: Regression rate. Panel d: Mean of individual re-reading fixations.
The baseline (bar) was calculated using data from the entire reading session. Error bars show
standard errors.
Figure 3. Time course of lexical and linguistic effects on gaze durations before and after new
song onset. Individualized regression model (Lorch & Meyers, 1990) was built for each bin as
well as for the entire reading session (bars). The plots showed unstandardized coefficients
averaged across participants. Error bars show standard errors.
... For example, Lim (2020) investigated the differences in cognitive processes between reading test-takers and reported that the participants' eye movements represented their bottom-up linguistic processing. Zhang et al. (2018) investigated whether reading with background music can affect reading comprehension. Results suggested that normal word recognition can still be executed in the presence of the background music and if the music causes disruption, participants compensate with the rereading of the passage. ...
... Since the listening comprehension process constitutes an interplay between the oral text, task requirements, and the listeners' neurocognitive processes (Rost 2015), inequalities in the features of the oral texts would affect listeners' processes (Révész and Brunfaut 2013) and, subsequently, their test scores. Generally, any change in the features of the input or the way it is presented could have some impact on not only the performance of test takers, but also their neurocognitive processes (e.g., Aryadoust 2019; Zhang et al. 2018). ...
The present study explored the potential of a new neurocognitive approach to test equity which integrates evidence from eye-tracking and functional near-infrared spectroscopy with conventional test content analysis and psychometric analysis. The participants of the study (n = 29) were neurotypical university students who took two tests of English lecture comprehension. Test equity was examined in this study at four levels: the linguistic level (content evidence) and the test scores level which are conventional levels in test equity; and gaze behavior level and neurocognitive level which are novel to this study. It was found that the linguistic features of the two test forms being equated were similar and that there was no significant difference at neurocognitive and behavioral levels. However, there was a significant difference in gaze behaviors, measured by fixation counts and visit counts, although fixation duration and visit duration did not vary across the two tests. Overall, test equity was supported, despite partial counterevidence from the gaze data. We discuss the implication of this approach for future equity research and response process in language assessment.
... Separating the overall effect into µ and τ can reflect different kinds of effects that MW could have on visual processing. It has been shown that, while some factors can change both µ and τ (e.g., word frequency; Staub et al., 2010), some only affect µ (e.g., lexical predictability; Staub, 2011) or τ (e.g., music distraction, Zhang, Miller, Cleveland, & Cortina, 2018). Simply looking at the aggregated measures might conceal these effects. ...
... An increase in the mu parameter is often seen in processing difficulty manipulations such as changes in word frequency and word predictability (Staub, 2011;Staub et al., 2010). On the other hand, an increase in the tau parameter was found with an increased level of external distraction (Zhang et al., 2018). Here, the results showed that an increase in the tau parameter could also be associated with internal distraction. ...
Mind-wandering (MW) is ubiquitous and is associated with reduced performance across a wide range of tasks. Recent studies have shown that MW can be related to changes in gaze parameters. In this dissertation, I explored the link between eye movements and MW in three different contexts that involve complex cognitive processing: visual search, scene perception, and reading comprehension. Study 1 examined how MW affects visual search performance, particularly the ability to suppress salient but irrelevant distractors during visual search. Study 2 used a scene encoding task to study how MW affects how eye movements change over time and their relationship with scene content. Study 3 examined how MW affects readers’ ability to detect semantic incongruities in the text and make necessary revisions of their understanding as they read jokes. All three studies showed that MW was associated with decreased task performance at the behavioral level (e.g., response time, recognition, and recall). Eye-tracking further showed that these behavioral costs can be traced to deficits in specific cognitive processes. The final chapter of this dissertation explored whether there are context-independent eye movement features of MW. MW manifests itself in different ways depending on task characteristics. In tasks that require extensive sampling of the stimuli (e.g., reading and scene viewing), MW was related to a global reduction in visual processing. But this was not the case for the search task, which involved speeded, simple visual processing. MW was instead related to increased looking time on the target after it was already located. MW affects the coupling between cognitive efforts and task demands, but the nature of this decoupling depends on the specific features of particular tasks.
... It is likely that multitasking diverted attention from reading the text, and readers cannot process a text without attending to it (Cowan, 1988;LaBerge & Samuels, 1974;Wu, 2017). To compensate for diverted attention, readers may be more likely to reread text after being distracted by multitasking (Zhang, Miller, Cleveland, & Cortina, 2018), which would explain the longer reading times. However, divided attention may interfere with memory as there is an interference effect in which information from one task is confused with information from the other task (Fernandes & Moscovitch, 2000;Pashler, 1994). ...
... One reason for this could be the physical distance involved between tasks (Jeong & Hwang, 2016). When reading from screens, the multitasking typically also involved the same device (e.g., Bowman et al., 2010;Kononova et al., 2016;Tran et al., 2013) or was an auditory stream of information that did not require physical movement (e.g., Zhang et al., 2018). In contrast, when reading from paper, the multitasking usually required the participant to move their head and eyes away from the paper (e.g., Armstrong & Chung, 2000;Fante et al., 2013;Jeong & Hwang, 2012). ...
Background: Multitasking while reading is a commonplace activity. Many studies have been conducted examining the effect of multitasking on reading comprehension and times. The purpose of this meta-analysis is to consolidate the empirical findings on reading comprehension and times in order to understand the overall effect of multitasking on reading. Characteristics of the reading situation, comprehension assessment, and the secondary task were examined to determine if they varied the effect of multitasking. Methods: A systematic search of studies on multitasking and reading was conducted. Only studies that used random assignment and had participants reading independently were included. This screening yielded a total of 22 independent studies (20 reports) that met inclusion criteria, with 20 studies on reading comprehension and 9 studies on reading times. Most of the studies involved adults reading expository texts. Results: Based on Robust Variance Estimation (RVE) analyses, multitasking had a negative effect on reading comprehension (g = -0.28, p = .002). The effect was similar after outliers were removed, (g = -0.26, p = .001). Based on moderator analyses, this negative effect may only occur when time was limited because the reading pace was controlled by the experimenter (g = -0.54, p < .001) as there was not a reliable effect when reading was self-paced (g = -0.14, p = .10). Multitasking during reading lead to longer reading times (g = 0.52, p < .001). Conclusions: Multitasking during reading is detrimental to reading comprehension when time is limited. When readers control their own pace of reading, multitasking lengthens the time for the reading task. Therefore, multitasking while reading is less efficient than focusing attention on the primary task of reading.
Web analytics has changed significantly in recent years. As part of the big data revolution, frequent low-level user actions, such as mouse movements and clicks, are often used in modern web analytics. Various studies show that when a user moves or clicks the mouse, the position of the mouse cursor is relatively close to the position of the eye gaze on the screen. Accordingly, mouse cursor positions can indicate user attention and interest in specific areas of web pages. This study focuses on mouse movement directions and speeds rather than on mouse cursor positions. A statistical analysis of mouse movements on an online learning website, which was selected for this study, sheds light on several interesting patterns. For example, most mouse movements in the examined usage data are either approximately horizontal or approximately vertical, and horizontal mouse movements are more frequent than vertical mouse movements. Besides, horizontal movements to the left are not equivalent to horizontal movements to the right, in terms of moving time and speed. As this study shows, these statistical findings are related to Pointer Assisted Reading (PAR), a reading behavior consisting of moving the mouse cursor (also known as the mouse pointer) along sentences, marking the reading position, similarly to finger-pointing when reading a book. Associating mouse movements with text reading may potentially highlight content that most users tend to skip, and therefore, might not interest the website's audience, as well as content that many readers read more than once or slowly, suggesting a lack of clarity or ambiguity. As discussed in this paper, this could be useful in locating issues in the textual content of websites and especially in online learning and educational technology applications.
Full-text available
La présence de sons non pertinents est reconnue pour affecter le fonctionnement cognitif. Plus précisément, la présentation d'un son qui dévie du contexte auditif a le potentiel d'affecter la performance à une tâche réalisée simultanément. Plusieurs études s'intéressant à l'impact délétère des sons déviants sur la performance à une tâche montrent qu'il découle de la réorientation de l'attention déclenchée lorsqu'un son déviant est détecté par l'organisme. Cette capture attentionnelle (ou réponse d'orientation) est également reconnue pour engendrer plusieurs réponses physiologiques associées à l'état d'alerte. Ces réponses physiologiques, sous certaines conditions, sont considérées comme des indices psychophysiologiques de la capture attentionnelle auditive permettant de démontrer l'occurrence d'une réorientation de l'attention de la tâche en cours vers le son déviant. De récentes études suggèrent qu'il existe un lien entre ces indices et la réponse pupillaire, soit l'augmentation rapide du diamètre pupillaire. Quelques études ont tenté d'évaluer si cette réponse pouvait remplir les critères d'un indice valide de la réponse d'orientation; cependant, les résultats de ces études sont parfois contradictoires ou incomplets. La présente thèse propose donc une évaluation systématique de l'utilisation de la réponse pupillaire à titre d'indice psychophysiologique de la capture attentionnelle auditive. Les résultats de la première étude montrent que la réponse pupillaire respecte les critères d'un index valide de la réponse d'orientation, supportant ainsi la validité de cet index. L'Étude 2 montre que cet indice peut être utilisé dans des contextes dans lesquels les sujets effectuent simultanément à l'écoute des sons une tâche visuelle provoquant des changements systématiques de luminosité ou des mouvements oculaires qui affectent la taille de la pupille. Ces expériences appuient ainsi l'utilisabilité de cet index. Enfin, la dernière étude permet d'établir l'utilité de ce proxy puisque ce dernier permet de distinguer l'origine de deux phénomènes de distraction auditive différents. Dans l'ensemble, les résultats de la thèse révèlent que la réponse pupillaire représente un indice psychophysiologique adéquat qui pourrait être intégré aux études sur la distraction auditive ou dans des contextes plus appliqués où les mesures de la capture attentionnelle auditive peuvent s'avérer pertinentes.
This study investigated how semantically relevant auditory information might affect the reading of subtitles, and if such effects might be modulated by the concurrent video content. Thirty-four native Chinese speakers with English as their second language watched video with English subtitles in six conditions defined by manipulating the nature of the audio (Chinese/L1 audio vs. English/L2 audio vs. no audio) and the presence versus absence of video content. Global eye-movement analyses showed that participants tended to rely less on subtitles with Chinese or English audio than without audio, and the effects of audio were more pronounced in the presence of video presentation. Lexical processing of subtitles was not modulated by the audio. However, Chinese audio, which presumably obviated the need to read the subtitles, resulted in more superficial post-lexical processing of the subtitles relative to either the English or no audio. On the contrary, English audio accentuated post-lexical processing of the subtitles compared with Chinese audio or no audio, indicating that participants might use English audio to support subtitle reading (or vice versa) and thus engaged in deeper processing of the subtitles. These findings suggest that, in multimodal reading situations, eye movements are not only controlled by processing difficulties associated with properties of words (e.g., their frequency and length) but also guided by metacognitive strategies involved in monitoring comprehension and its online modulation by different information sources.
Full-text available
This paper explores Pointer Assisted Reading (PAR), a reading behavior consisting of moving the mouse cursor (also known as the pointer) along sentences to mark the reading position, similarly to finger-pointing when reading a book. The study shows that PAR is an uncommon reading technique and examines methods to extract and visualize the PAR activity of web users. An analysis shows that PAR data of real users reveal reading properties, such as speed, and reading patterns, such as skipping and rereading. Eye-tracking is usually used to analyze user reading behaviors. This paper advocates for considering PAR-tracking as a feasible alternative to eye-tracking on websites, as tracking the eye gaze of ordinary web users is usually impractical. PAR data might help in spotting quality issues in the textual content of a website, such as unclear text or content that might not interest the website users, based on analyzing reading properties and patterns (e.g. reading speed, skipping, and rereading). Accordingly, PAR-tracking may have various practical applications in a wide range of fields, and particularly in educational technology, e-learning, and web analytics.
Full-text available
The effects of background speech or noise on visually based cognitive tasks has been widely investigated; however, little is known about how the brain works during such cognitive tasks when music, having a powerful function of evoking emotions, is used as the background sound. The present study used event-related potentials to examine the effects of background music on neural responses during reading comprehension and their modulation by musical arousal. Thirty-nine postgraduates judged the correctness of sentences about world knowledge without or with background music (high-arousal music and low-arousal music). The participants' arousal levels were reported during the experiment. The results showed that the N400 effect, elicited by world knowledge violations versus correct controls, was significantly smaller for silence than those for high-and low-arousal music backgrounds, with no significant difference between the two musical backgrounds. This outcome might have occurred because the arousal levels of the participants were not affected by the high-and low-arousal music throughout the experiment. These findings suggest that background music affects neural responses during reading comprehension by increasing the difficulty of semantic integration, and thus extend the irrelevant sound effect to suggest that the neural processing of visually based cognitive tasks can also be affected by music.
Full-text available
In the present study we measured the eye movements of a large sample of 2(nd) grade German speaking children and a control group of adults during a silent reading task. To be able to directly investigate the interaction of word length and frequency effects we employed controlled sentence frames with embedded target words in an experimental design in which length and frequency were manipulated independently of one another. Unlike previous studies which have investigated the interaction of word length and frequency effects in children, we used age-appropriate word frequencies for children. We found significant effects of word length and frequency for both children and adults while effects were generally greater for children. The interaction of word length and frequency was significant for children in gaze duration and total viewing time eye movement measures but not for adults. Our results suggest that children rely on sublexical decoding of infrequent words, leading to greater length effects for infrequent than frequent words while adults do not show this effect when reading children's reading materials. Copyright © 2015. Published by Elsevier Ltd.
Full-text available
The study examined the nature of eye movement control and word recognition during scanning for a specific topic, compared with reading for comprehension. Experimental trials included a manipulation of word frequency: the critical word was frequent (and orthographically familiar) or infrequent (2 conditions: orthographically familiar and orthographically unfamiliar). First-pass reading times showed effects of word frequency for both reading and scanning, with no interactions between word characteristics and task. Therefore, in contrast to the task of searching for a single specific word (Rayner & Fischer, 1996), there were immediate and localized influences of lexical processing when scanning for a specific topic, indicating that early word recognition processes are similar during reading and topic scanning. In contrast, there were interactions for later measures, with larger effects of word frequency during reading than scanning, indicating that reading goals can modulate later processes such as the integration of words into sentence context. Additional analyses of the distribution of first-pass single fixation durations indicated that first-pass fixations of all durations were shortened during scanning compared with reading, and reading for comprehension produced a larger subset of longer first-pass fixations compared with scanning. The implications for the nature of word recognition and eye movement control are discussed. (PsycINFO Database Record (c) 2014 APA, all rights reserved).
This article presents the Provo Corpus, a corpus of eye-tracking data with accompanying predictability norms. The predictability norms for the Provo Corpus differ from those of other corpora. In addition to traditional cloze scores that estimate the predictability of the full orthographic form of each word, the Provo Corpus also includes measures of the predictability of the morpho-syntactic and semantic information for each word. This makes the Provo Corpus ideal for studying predictive processes in reading. Some analyses using these data have previously been reported elsewhere (Luke & Christianson, 2016). The Provo Corpus is available for download on the Open Science Framework, at .
Several studies have examined effects of explicit task demands on eye movements in reading. However, there is relatively little prior research investigating the influence of implicit processing demands. In the present study, processing demands were manipulated by means of a between-subject manipulation of comprehension question difficulty. Consistent with previous results from Wotschack and Kliegl (2013), the question difficulty manipulation influenced the probability of regressing from late in sentences and re-reading earlier regions; readers who expected difficult comprehension questions were more likely to re-read. However, this manipulation had no reliable influence on eye movements during first pass reading of earlier sentence regions. Moreover, for the subset of sentences that contained a plausibility manipulation, the disruption induced by implausibility was not modulated by the question manipulation. We interpret these results as suggesting that comprehension demands influence reading behavior primarily by modulating a criterion for comprehension that readers apply after completing first-pass processing.
This chapter explores the range of consciousness occurring within the everyday music experiences of a small sample of UK listeners, particularly those experiences lying between the extremes of intense, emotional involvement, and apparent inattention when music, though present, seems to be barely perceived. Specifically, it draws on the constructs of trance, absorption, and dissociation as explicatory frames that throw into relief the self-regulating character - in psychological terms - of much everyday listening. By concentrating on the detailed nature of music listening episodes as lived experiences it becomes possible to offer a phenomenology of everyday listening, thus 'reclaiming' it for comparison with the literature on strong experiences.
Dissociative experiences involving music have received little research attention outside the field of ethnomusicology. This paper examines the psychological characteristics of normative dissociation (detachment) across musical and non-musical experiences in ‘real world’, everyday settings. It draws upon a subset of data arising from an empirical project designed to compare transformative shifts of consciousness, with and without music in daily life, and the ways in which use of music may facilitate the processes of dissociation and absorption. Twenty participants kept unstructured diaries for two weeks, recording free descriptions of involving experiences of any kind as soon as possible after their occurrence. All descriptions were subsequently subjected to Interpretative Phenomenological Analysis (IPA). Results suggest that dissociative experiences are a familiar occurrence in everyday life. Diary entries highlight an established practice of actively sought detachment from self, surroundings or activity, suggesting that, together with absorption, the processes of derealization (altered perception of surroundings) and depersonalization (detachment from self) constitute common means of self-regulation in daily life. Music emerges as a particularly versatile facilitator of dissociative experience because of its semantic ambiguity, portability, and the variety of ways in which it may mediate perception, so facilitating an altered relationship to self and environment.
Music listening in everyday life tends to accompany the completion of other everyday activities in a highly personalised manner. However, music and task performance studies have tended to be experimenter-centred and contextually isolated, largely independent of the listener's music practices and preference. The present study adopted a listener-centred approach to compare the effects of self-selected and experimenter-selected music (high and low arousal), on concurrent activity performance and experience. 125 participants completed three laps of a driving game in either (i) silence (ii) car sounds alone; car sounds with the addition of (iii) self-selected music, (iv) High-Arousal music or (v) Low-Arousal music. Three performance measures (accuracy-collisions, time-ms, and speed-mph) and 5 experience measures (distraction, liking, appropriateness, enjoyment, and tension-anxiety) were taken. Participants exposed to their self-selected music were most efficient, perceived lowest distraction, highest enjoyment, liking and appropriateness, and experienced a reduction in tension-anxiety. In contrast, performance and experience were poorest when exposed to High-Arousal experimenter-selected music. Participants were most inaccurate, perceived highest distraction, lowest liking, enjoyment and appropriateness, and experienced an increase in tension-anxiety. Collectively, the findings highlight the efficacy of self-selected music as a tool to optimise response in the everyday activity context for which it is selected. Accordingly, the results are discussed in relation to potential implications for the performance and experience of concurrent tasks such as video games. Additionally, the discussion highlights theories of attention-distraction, arousal and affect modification, and subjective experiences of music listening.