ArticlePDF Available

Semantic processing of unattended speech in dichotic listening

Authors:

Abstract

This study investigated whether unattended speech is processed at a semantic level in dichotic listening using a semantic priming paradigm. A lexical decision task was administered in which target words were presented in the attended auditory channel, preceded by two prime words presented simultaneously in the attended and unattended channels, respectively. Both attended and unattended primes were either semantically related or unrelated to the attended targets. Attended prime-target pairs were presented in isolation, whereas unattended primes were presented in the context of a series of rapidly presented words. The fundamental frequency of the attended stimuli was increased by 40 Hz relative to the unattended stimuli, and the unattended stimuli were attenuated by 12 dB [+12 dB signal-to-noise ratio (SNR)] or presented at the same intensity level as the attended stimuli (0 dB SNR). The results revealed robust semantic priming of attended targets by attended primes at both the +12 and 0 dB SNRs. However, semantic priming by unattended primes emerged only at the 0 dB SNR. These findings suggest that the semantic processing of unattended speech in dichotic listening depends critically on the relative intensities of the attended and competing signals.
!
1
The following article appeared in the Journal of the Acoustical Society of America and may be found at
(http://dx.doi.org/10.1121/1.4927410). Citation: Aydelott, J., Jamaluddin, Z., & Pearce, S. N. (2015)
Semantic processing of unattended speech in dichotic listening. Journal of the Acoustical Society of
America, 138(2), 964975.
Copyright (2015) Acoustical Society of America. This article may be downloaded for personal use only.
Any other use requires prior permission of the author and the Acoustical Society of America.
Semantic processing of unattended speech
in dichotic listening
Jennifer Aydelott, Zahra Jamaluddin, and Stefanie Nixon Pearce
Department of Psychological Sciences, Birkbeck, University of London
Please address correspondence to:
Dr Jennifer Aydelott, Department of Psychological Sciences, Birkbeck, University of London,
Malet Street, London WC1E 7HX, United Kingdom.
Email: j.aydelott@bbk.ac.uk
Running title: SEMANTIC PROCESSING OF UNATTENDED SPEECH
!
2
ABSTRACT
This study investigated whether unattended speech is processed at a semantic level in
dichotic listening using a semantic priming paradigm. A lexical decision task was administered
in which target words were presented in the attended auditory channel, preceded by two prime
words presented simultaneously in the attended and unattended channels, respectively. Both
attended and unattended primes were either semantically related or unrelated to the attended
targets. Attended prime–target pairs were presented in isolation, whereas unattended primes were
presented in the context of a series of rapidly presented words. The fundamental frequency of the
attended stimuli was increased by 40 Hz relative to the unattended stimuli, and the unattended
stimuli were attenuated by 12 dB (+12 dB signal-to-noise ratio/SNR) or presented at the same
intensity level as the attended stimuli (0 dB SNR). The results revealed robust semantic priming
of attended targets by attended primes at both the +12 and 0 dB SNRs. However, semantic
priming by unattended primes emerged only at the 0 dB SNR. These findings suggest that the
semantic processing of unattended speech in dichotic listening depends critically on the relative
intensities of the attended and competing signals.
PACS numbers: 43.71.Sy, 43.66.Rq, 43.71.Es, 43.66.Lj
!
3
I. INTRODUCTION
In a multitalker environment such as a cocktail party, listeners are able to compensate for
the distracting influence of competing voices by focusing attention on a single talker (Cherry,
1953; Broadbent, 1958). Selective attention to spatial, temporal, and spectral cues that
distinguish the attended signal from other information in the auditory scene improves
identification thresholds for attended speech (see Bronkhorst, 2000, for review). The benefit of
spatial separation for the segregation of competing speech streams has been exploited in studies
of dichotic listening, in which attended and competing signals are presented to separate auditory
channels. In the classic dichotic listening paradigm, listeners are able to repeat spoken words
presented to the attended channel while ignoring competing speech presented to the contralateral
ear, without apparent awareness of the content of the competing message (Cherry, 1953;
Broadbent, 1958; Moray, 1959; cf. Wood & Cowan, 1995a). However, although dichotic
presentation facilitates selective attention to the relevant channel, listeners are able to detect
changes in the acoustic characteristics of the unattended signal, e.g., a shift from one voice to
another (Treisman & Riley, 1969). Further, meaningful information from the competing signal
can capture attention away from the attended signal if it is salient to the listener (e.g., a listener’s
own name; Moray, 1959; Wood & Cowan, 1995b) or closely related to the attended message
(e.g., Gray & Wedderburn, 1960; Treisman, 1960; Mackay, 1973; see Holender, 1986, for an
extensive review). Thus, under certain conditions, listeners are sensitive to the content of the
competing speech signal even when attention is focused away from the irrelevant channel.
The intrusion of an unattended speech signal can have adverse effects on the
identification and comprehension of attended speech. The meaningful content of a competing
signal can contribute to informational masking effects (cf. Schneider, Li, & Daneman, 2007),
!
4
such that poorer speech recognition performance is observed when a speech masker is in a
listener’s native language relative to a foreign language (Garcia Lecumberri & Cooke, 2006; Van
Engen & Bradlow, 2007), or when a masker contains high-frequency words relative to low-
frequency words (Boulenger et al., 2010). Further, meaningful competing speech can disrupt the
semantic processing of attended sentences under moderately demanding listening conditions. At
a signal-to-noise ratio (SNR) of 0 dB, listeners show reduced sensitivity to sentence-level
semantic cues in dichotic listening when meaningful speech is presented in the unattended
channel, but not when the same speech signal is played backward (Aydelott et al., 2012). The
conditions under which listeners access higher-level linguistic information from an unattended
speech signal are therefore critical to our understanding of spoken language comprehension in a
multitalker environment.
Recent studies have used implicit priming measures to investigate lexical processing of
unattended speech in dichotic listening. In a series of experiments, Dupoux, Kouider, and Mehler
(2003) examined whether spoken words presented in an unattended channel would influence
reaction times (RTs) to the same words presented in the attended channel in a lexical decision
task. Unattended primes were time-compressed to 38% of their original duration and presented at
an intensity level of –12 dB relative to the attended stimuli (+12 dB SNR). When unattended
primes and attended targets were presented simultaneously, identity priming emerged in response
to primes presented in isolation, but not to primes embedded in a sentence context. In contrast,
when unattended primes immediately preceded attended targets, identity priming was observed
in response to primes embedded in a sentence context, but not to primes embedded in a
multitalker babble. Dupoux et al. (2003) attribute this pattern of effects to the relative salience of
the prime stimuli across conditions: the abrupt onset of a prime presented in isolation is more
!
5
likely to draw attention than a prime in continuous speech when the prime coincides with the
target, whereas the prosodic structure of a sentence context is more likely than unstructured
babble noise to draw attention to the prime when it occurs before the target. On the basis of these
findings, Dupoux et al. (2003) concluded that unattended words are processed at the lexical level
only when they are sufficiently salient to capture attention away from the attended channel.
This claim was challenged by Rivenez, Darwin, and Guillaume (2006), however, who
argued that the physical characteristics of the stimuli used by Dupoux et al. (2003) may have led
to an underestimation of the identity priming effect for unattended primes. According to Rivenez
et al. (2006), the acoustic degradation introduced by temporal compression of the prime stimuli
in the Dupoux et al. (2003) study, in combination with the demands on auditory segregation
imposed by the use of the same voice for both primes and targets, may have reduced lexical
activation for unattended prime words. Further, the unattended prime stimuli in the Dupoux et al.
(2003) study were presented exclusively to the left auditory channel, which is at a disadvantage
relative to the right channel in the processing of speech stimuli due to the preponderance of
neural projections to the right hemisphere of the brain (Kimura, 1961, 1967; cf. Hugdahl, 2005).
To address these issues, Rivenez et al. (2006) used a modified version of Dupoux et al.’s
(2003) paradigm to examine whether lexical priming would emerge for unattended stimuli in
dichotic listening under conditions more favorable to perceptual grouping. A continuous stream
of words was presented in the attended channel, and sentences were presented in the unattended
channel at an intensity level of –12 dB relative to the attended stimuli (+12 dB SNR). The
fundamental frequency (F0) of the attended words was flattened and increased by 40 Hz relative
to the unattended sentences so that the attended and unattended signals were in different pitch
ranges, and the ear of presentation of the attended words was counterbalanced so that responses
!
6
to right-channel and left-channel unattended speech could be compared. Listeners were asked to
monitor the attended word stream and respond upon detection of targets belonging to a particular
semantic category. Prime words presented in the unattended channel consisted of a token of the
target word produced in the context of the unattended sentence, presented without temporal
compression immediately before the onset of the target in the attended channel. Rivenez et al.
(2006) found significant facilitation of targets preceded by unattended primes under these
conditions, even when attentional switches were controlled using a secondary task. Moreover,
the priming effect was not influenced by the ear of presentation of the prime stimuli. By contrast,
no priming was observed when the attended and competing signals were presented at the same
F0.
The findings of Rivenez et al. (2006) suggest that unattended stimuli can influence the
processing of attended stimuli even in the absence of a shift of attention to the irrelevant channel.
Further, these results demonstrate that the acoustic conditions under which competing signals are
presented play a key role in determining whether unattended speech is subject to higher-level
language processing. Subsequent studies have confirmed that implicit identity priming of
attended words by unattended words embedded in continuous speech may be observed when the
two speech signals can be differentiated by talker characteristics such as vocal tract length
(Rivenez, Darwin, Bourgeon, & Guillaume, 2007) or a combination of pitch and timbre (Rivenez,
Guillaume, Bourgeon, & Darwin, 2008). Rivenez et al. (2007) propose that perceptual grouping
is a necessary prerequisite for the lexical processing of unattended speech.
Although the work of Rivenez and colleagues indicates that unattended words can
produce lexical priming under conditions that facilitate auditory segregation (i.e., perceptual
grouping of concurrent sounds into separate auditory streams), it remains to be established
!
7
whether unattended speech stimuli are processed at the semantic level. A previous study by
Bentin, Kutas, and Hillyard (1995) used event-related brain potentials (ERPs) and behavioural
measures to examine semantic priming effects in dichotically presented stimuli. Two lists of
isolated spoken words were presented simultaneously to separate auditory channels, and words
in the attended and unattended lists were either semantically related or unrelated to subsequent
words in the attended list. The ERP data showed an N400 semantic priming effect for attended
primes, but not for unattended primes. However, a follow-up recognition memory task revealed
an increased rate of false positive responses to novel words that were semantically related to the
unattended primes, providing indirect evidence that the unattended stimuli were subject to
implicit semantic processing. Nevertheless, because the stimuli were presented as isolated words
separated by silence, this finding may reflect inadvertent shifts of attention to the unattended
channel, as suggested by Dupoux et al. (2003).
By contrast, in both the lexical decision paradigm used by Dupoux et al. (2003) and the
semantic detection task used by Rivenez et al. (2006), the prime and target were different tokens
of the same word, so the observed priming effects may be attributable to the activation of a
shared lexical form representation. In the case of Rivenez et al. (2006), although the detection
task required listeners to identify targets based on semantic category membership, the results can
be accounted for in terms of form-based priming, which would facilitate target recognition in the
attended stream, resulting in earlier semantic analysis and faster detection of appropriate targets.
Behavioural studies of lexical priming in dichotic listening have not examined whether
facilitation of attended targets by unattended primes can be obtained on the basis of a semantic
relationship between the prime and target. Further, previous investigations have not explored the
relative effects of information in the attended channel on the processing of attended targets,
!
8
which have been presented either in isolation (Dupoux et al., 2003) or preceded by unrelated
material (Rivenez et al., 2006, 2007). Thus, these studies do not allow for an examination of the
simultaneous effects of attended and unattended speech on subsequent responses to attended
words.
The present study used a semantic priming paradigm to assess implicit semantic
processing of unattended words in dichotic listening. Prime words embedded in a continuous
speech stream were presented in the unattended auditory channel to examine their effect on
target words presented in the attended channel. In addition, prime words were also presented in
the attended channel at the same time as the unattended primes, to provide a simultaneous
measure of the semantic processing of attended and unattended stimuli. The attended and
competing prime words were either semantically related or unrelated to the attended targets, and
there was no phonological relationship between primes and targets. As in Rivenez et al. (2006),
the F0 of the attended stimuli was increased by 40 Hz relative to the unattended stimuli to
facilitate perceptual grouping of the two speech streams, and the ear of presentation of the
attended stimuli was counterbalanced to control for the right-ear advantage (REA) for speech
processing (Kimura, 1961, 1967). However, the design of the present study differed from
Rivenez et al. (2006) in several respects. As Rivenez et al. (2006) observe, the go/no-go semantic
detection task used in their study was highly demanding in terms of sustained attention. Both the
attended and unattended signals in Rivenez et al. (2006) were continuous speech streams, and the
attended stream consisted of an uninterrupted series of pitch-flattened words, whereas the
competing stream consisted of sentences with a varying intonation contour. Listeners were
required to monitor the attended stream throughout the experiment while ignoring the more
salient unattended stream, which also provided prosodic and grammatical cues to the location of
!
9
the prime stimuli. Thus, the likelihood of attentional shifts to the unattended channel in this
paradigm was high, and had to be controlled using a secondary task. These issues were addressed
in the present study as follows: (1) attended stimuli were presented as isolated prime–target pairs,
rather than a list of words; (2) unattended primes were embedded in a series of rapidly presented
words, rather than a sentence context; and (3) listeners made a lexical decision judgment in
response to each attended target, rather than monitoring the attended channel for intermittent
targets. Attended primes in the present study therefore provided a cue to the timing of the target,
and served to orient the listener toward the attended channel and away from the unattended
channel. In contrast, unattended primes were likely to be less acoustically salient than the
isolated attended primes, as they were presented in the context of a continuous speech stream
that did not provide acoustic or linguistic cues to the timing of the prime stimuli (cf. Dupoux et
al., 2003; Rivenez et al., 2008).
Experiment 1 served as a norming study to establish the semantic priming effect
produced by simultaneous, dichotically presented prime stimuli in a traditional paired priming
paradigm under conditions of divided attention. In Experiment 2, the same stimuli were
presented in a selective attention paradigm in which primes in the unattended channel were
embedded in a continuous speech stream to examine whether the unattended speech stimuli
would be subject to semantic processing when the attended and competing speech streams
differed in F0. As in Dupoux et al. (2003) and Rivenez et al. (2006), the attended and unattended
stimuli in Experiment 2 were presented at a SNR of +12 dB. We predicted that attended primes
would produce semantic priming of attended targets in Experiment 2, i.e., that RTs would be
faster in response to targets preceded by semantically related primes than semantically unrelated
primes in the attended channel. We further predicted that, if unattended stimuli are processed at
!
10
the semantic level under conditions conducive to perceptual grouping (in this case, attended and
competing speech streams differing in F0 presented to separate auditory channels), competing
primes would also produce semantic priming of attended targets in Experiment 2. Experiment 3
tested the effects of more demanding listening conditions on the semantic processing of
unattended speech by examining semantic priming effects produced by attended and competing
primes when the relative intensity of the two signals was adjusted to a SNR of 0 dB.
II. EXPERIMENT 1
The purpose of Experiment 1 was to develop a set of prime and target stimuli to test the
hypotheses outlined above. As the effects of two simultaneous, dichotically presented spoken
prime words on responses to semantically related and unrelated target words have not been
explored in previous studies, Experiment 1 was designed to establish a baseline dichotic
semantic priming effect under divided attention conditions. Dichotic primes consisted of two
isolated words presented simultaneously to separate auditory channels in a paired prime–target
paradigm. Prime words in each auditory channel were either related or unrelated to the target
words. Listeners were instructed to listen to both primes and make a lexical decision response to
the target item. As the goal of Experiment 1 was to obtain a measure of dichotic priming in the
absence of selective attention, targets were presented diotically to avoid introducing an
attentional bias in favour of one channel and to ensure that lexical decision judgments would be
based on auditory information from both channels. All primes in one channel were presented at
an intensity level of –12 dB relative to the primes in the other channel to reflect the SNR used in
previous studies of dichotic repetition priming (Dupoux et al., 2003; Rivenez et al., 2006). Both
sets of dichotic prime stimuli were expected to produce semantic priming of related targets, i.e.,
!
11
RTs were expected to be faster to targets preceded by related primes than by unrelated primes.
However, to address the possibility that the simultaneous presentation of two competing primes
might interfere with semantic activation, thereby eliminating the semantic priming effect, an
additional condition was included in which prime stimuli in one channel were presented in
isolation without competing primes in the other channel.
A. Method
1. Stimuli
Sixty one-syllable words with an average length of three phonemes (mean = 3.27, SD =
0.64) were selected from the MRC Psycholinguistic Database (Coltheart, 1981) to serve as
targets in the lexical decision task. Target words had a mean Zipf frequency value of 4.97 (SD =
0.31) according to the SUBTLEX-UK database (Van Heuven, Mandera, Keuleers, & Brysbaert,
2014), indicating medium to high frequency of occurrence in British English. Each target word
was matched with two one-syllable words that were semantically related to the target, and two
semantically unrelated one-syllable words. One related and one unrelated word were selected to
serve as attended primes in Experiment 2, and the remaining words served as competing primes
in Experiment 2, for a total of 240 prime words. As these prime stimuli were presented under
divided attention conditions, neither set of primes was attended or competing in Experiment 1;
however, to clarify the relationship between the stimulus conditions in the two experiments, the
method and results of Experiment 1 are reported in terms of the conditions in which the prime
stimuli were to be presented in Experiment 2, i.e., “attended” and “competing” primes. Length,
frequency, and word association statistics for these stimuli are shown in Table I. Two-way
repeated measures analyses of variance (ANOVAs) revealed no significant differences in
SUBTLEX-UK Zipf frequency value between “attended” and “competing” primes, F(1,59) =
!
12
0.35, p = .556, or between related and unrelated primes, F(1,59) = 1.96, p = .166, with no
significant interaction of these factors, F(1,59) = 0.08, p = .782. For semantically related prime
stimuli, the proportion of target word responses elicited by the prime in a word association task
(Edinburgh Associative Thesaurus; Kiss, Armstrong, Milroy, & Piper, 1973) did not differ
significantly between “attended” and “competing” primes, paired t(59) = 1.57, p = .122. A
complete list of prime and target stimuli is provided in the Appendix.
Sixty one-syllable nonwords served as distractor targets in the lexical decision task. All
nonwords were phonologically permissible and did not differ significantly from the target words
in length (number of phonemes: mean = 3.35, SD = 0.58), independent t(118) = –0.45, p = .656.
To match the experimental conditions used for the test items, each nonword was paired with four
words that served as “attended” and “competing” primes. Two of these words were semantically
unrelated to the other primes, and two were semantically related to each other. The 240 distractor
primes were an average of 3.49 phonemes in length (SD = 0.74) with a mean SUBTLEX-UK
Zipf frequency value of 4.12 (SD = 0.73).
All stimulus words were read aloud by a male native speaker of Southern British English,
who was instructed to maintain a consistent rate and pitch. The stimuli were recorded onto digital
audiotape in an audiometric chamber with a Tascam DA-P1 tape recorder and a Sennheiser
ME65/K6 supercardioid microphone and pre-amp. The recording was transferred via digital-to-
digital sampling onto a Macintosh G4 computer with a Digidesign MBox using ProTools LE
software at a sampling rate of 44.1 kHz with a 16-bit quantization. The waveform of each item
was excised with onset and offset cuts made at zero-crossings, and saved in its own mono audio
file in WAV format for subsequent manipulation using Praat software (Boersma & Weenink,
2010). For each set of four prime words to be paired with a target word, the average duration of
!
13
the four waveforms was calculated, and each of the four waveforms was scaled to this duration
using the PSOLA algorithm in Praat; thus, all dichotically presented prime stimuli were of the
same duration.
“Attended” primes and target words and nonwords were scaled to the same nominal
mean rms amplitude in Praat. “Competing” primes were scaled to a nominal mean rms amplitude
of –12 dB relative to the “attended” primes and targets, consistent with the relative amplitudes of
attended and competing primes in Dupoux et al. (2003) and Rivenez et al. (2006). Prime stimuli
were generated by creating a stereo sound file in Praat with the waveform of the “attended”
prime word in the left channel, and the waveform of the “competing” prime word or silence in
the right channel. Six dichotic prime stimuli were produced for each target item, reflecting all
possible combinations of “attended” and “competing” primes (related/related, related/unrelated,
unrelated/related, unrelated/unrelated) as well as “attended” primes presented in isolation
(related/none, unrelated/none).
2. Procedure
Prime and target stimuli were presented to 48 native speakers of British English in a
lexical decision paradigm using a Macintosh iBook G4, Sennheiser HD25 headphones, and a
Cedrus RB-610 box, at a comfortable listening level. Prime–target pairs were presented with an
inter-stimulus interval of 500 msec and an inter-trial interval of 1000 msec. Primes were
presented dichotically and targets were presented diotically. Ear of presentation was not treated
as a separate variable; however, for 22 participants the stimuli were presented with the reversible
headphones in their standard left-right orientation, and for the remaining 26 participants the
headphones were reversed. Conditions were rotated across participants so that each prime and
target was presented only once for each participant (see also Experiment 2, below). RTs were
!
14
recorded using SuperLab Pro 1.75 software, and were measured from the onset of the target
stimuli. Testing was conducted in a sound-treated testing room or a suitable quiet location in the
field. Participants were instructed to listen to the prime and target stimuli, and to indicate
whether each target was a real word in English or not by pressing one of two buttons on the
response box. Participants were not instructed to attend selectively to one auditory channel; thus,
the results of Experiment 1 reflect the magnitude of priming obtained under divided attention
conditions. However, as noted above, the results are reported in terms of the conditions in which
the prime stimuli were to be presented in Experiment 2, i.e., “attended” and “competing” primes.
B. Results
Accuracy (percent correct) and harmonic mean RTs for correct responses were calculated
for each participant in each condition. The harmonic mean is based on the inverse transformation
and was used to minimize the influence of outliers and normalize the distribution of the RT data
(Ratcliff, 1993). Repeated measures ANOVAs were conducted on the accuracy and RT data,
with “Attended” Prime (related or unrelated) and “Competing” Prime (none, related, or
unrelated) as within-subjects variables. The results of the accuracy analysis revealed a significant
main effect of “Attended” Prime, F(1,47) = 13.16, p < .0001, partial η2 = .219, such that
participants were more accurate in responding to targets preceded by a related prime (mean %
correct = 98.06, SD = 4.12) than an unrelated prime (mean % correct = 95.14, SD = 8.53). The
main effect of “Competing” Prime was not significant, F(1,47) = 0.97, p = .383, partial η2 = .020,
and there was no significant interaction of the “Attended” and “Competing” prime conditions,
F(2,46) = 1.75, p = .180, partial η2 = .036. Thus, whereas a significant semantic priming effect
emerged in the accuracy analysis for “Attended” primes, “Competing” primes did not influence
!
15
the accuracy of lexical decision responses (related “Competing” primes, mean % correct = 96.77,
SD = 5.10; unrelated “Competing” primes, mean % correct = 96.04, SD = 7.78; no “Competing”
prime, mean % correct = 96.98, SD = 6.98). However, as overall accuracy was near ceiling
(grand mean > 96% correct), this measure may not have been sufficiently sensitive to detect
semantic facilitation by “Competing” primes, which were presented at a lower intensity level
than “Attended” primes.
Analyses of harmonic mean RTs for correct responses to word targets revealed a
significant main effect of “Attended” Prime, F(1,47) = 20.08, p < .0001, partial η2 = .299, such
that RTs were faster to targets preceded by related “Attended” primes (mean = 1182, SD = 217)
than unrelated “Attended” primes (mean = 1226, SD = 216). A significant main effect of
“Competing” Prime was also observed, F(2,46) = 13.40, p < .0001, partial η2 = .368. Paired t-
tests revealed significantly faster RTs to targets preceded by related “Competing” primes (mean
= 1166, SD = 195) than by unrelated “Competing” primes (mean = 1211, SD = 240), t(47) = –
3.35, p = .002, or by “Attended” primes without a “Competing” prime (mean = 1235, SD = 223),
t(47) = –5.22, p < .0001. There was no significant interaction of the “Attended” and “Competing”
prime conditions, F(2,46) = 0.12, p = .886, partial η2 = .005. Thus, in the RT analysis, highly
significant semantic priming effects emerged for both the “Attended” prime stimuli (44 msec)
and the “Competing” prime stimuli (45 msec).
C. Discussion
The results of Experiment 1 demonstrate dichotic semantic priming in a paired prime–
target paradigm under conditions of divided attention. When two spoken prime words were
presented simultaneously to separate auditory channels at a relative intensity level of +12 dB,
!
16
both primes produced faster RTs to semantically related targets than to unrelated targets. Thus,
semantic facilitation of related target words was observed for each of two simultaneous
competing primes, indicating that both primes produced semantic activation.
These stimuli were used to investigate semantic priming effects under selective attention
conditions in Experiment 2. Unlike Experiment 1, listeners in Experiment 2 were instructed to
listen to one channel and ignore the other. “Attended” primes and target items were presented in
pairs in the attended channel, and “competing” primes were embedded in a continuous stream of
words and presented simultaneously with “attended” primes in the unattended channel. Attended
and competing stimuli were presented at SNR of +12 dB, i.e., the attended stimuli (signal) were
presented at an intensity level of +12 dB relative to the competing stimuli (noise). To facilitate
perceptual grouping of the two speech streams, the F0 of the attended stimuli was increased by
40 Hz (Rivenez et al., 2006). Repetition priming has been demonstrated for primes in the
unattended channel under similar conditions, indicating that unattended prime stimuli are
processed at the lexical level (Rivenez et al., 2006). Experiment 2 examined whether unattended
primes presented under these conditions would produce activation at the semantic level.
III. EXPERIMENT 2
A. Method
1. Stimuli
The same prime and target stimuli from Experiment 1 were used in Experiment 2.
“Attended” prime stimuli from Experiment 1 served as attended primes and “competing” prime
stimuli served as competing primes. An additional set of 280 one-syllable words served as
distractor stimuli, which were presented as a continuous word stream. Distractor words were an
!
17
average of 3.35 phonemes in length (SD = 0.72) with a mean SUBTLEX-UK Zipf frequency
value of 4.13 (SD = 0.77). All distractor stimuli were produced by the same talker and recorded
and digitized in the same manner as the prime and target stimuli.
Attended primes and target words and nonwords were scaled to the same nominal mean
rms amplitude in Praat. Competing primes and distractor words were scaled to a nominal mean
rms amplitude of –12 dB relative to the attended primes and targets, yielding a SNR of +12 dB.
The pitch of the attended primes and targets was increased by 40 Hz in Praat using PSOLA. Four
dichotic prime stimuli were produced for each target item, reflecting all possible combinations of
attended and competing primes (related/related, related/unrelated, unrelated/related, and
unrelated/unrelated).
A continuous word stream was generated by creating a stereo sound file and inserting the
waveforms of distractor words sequentially into the right channel. Each of the 280 distractor
words was repeated three times over the course of the experiment. Distractor words were
selected so that the two items immediately preceding and following the prime–target pair in each
trial had no semantic or phonological relationship to the primes or the target. Two sets of word-
stream stimuli were produced: a lead-in set to be presented before the prime stimuli, and a
terminal set to be presented with the target stimuli. Each lead-in word-stream stimulus consisted
of two distractor words, and each terminal word-stream stimulus consisted of five distractor
words. No distractor word was repeated within a trial, and no distractor word served as both the
first item in a lead-in stimulus and the last item in a terminal stimulus. Target words and
nonwords were inserted into the left channel at the onset of the terminal word-stream stimuli.
!
18
2. Procedure
Each trial consisted of a lead-in word-stream stimulus, a prime stimulus, and a
target/terminal word-stream stimulus presented sequentially with a 0 msec inter-stimulus interval,
followed immediately by the next trial with a 0 msec inter-trial interval. Thus, stimuli in the
unattended channel were presented as a continuous stream of words, and stimuli in the attended
channel were presented as intermittent prime–target pairs. This trial sequence is illustrated in
Figure 1. Targets were presented in four prime conditions: related attended prime with related
competing prime (related/related); related attended prime with unrelated competing prime
(related/unrelated); unrelated attended prime with related competing prime (unrelated/related);
and unrelated attended prime with unrelated competing prime (unrelated/unrelated). All
participants received the lead-in word-stream and target/terminal word-stream stimuli. Prime
conditions were rotated such that all participants received all prime conditions, but the targets
that were presented in each prime condition were counterbalanced across participants, and each
prime and target stimulus was presented only once for each participant. Thus, all target items
were presented in all conditions, and all participants received all target items, thereby controlling
for item-specific effects (McNamara, 2005).
The experiment was conducted on a Macintosh PowerPC G4 eMac computer using
SuperLab software. All auditory stimuli were presented through Sennheiser HD 25-1 headphones
in a sound-treated testing room. A comfortable presentation level was established a priori by the
experimenters, and this level was held constant for all participants. Ear of presentation of the
attended prime and target was determined for each participant by the placement of the reversible
headphones. RTs and accuracy were recorded in SuperLab using a Cedrus RB-730 response box.
RTs were measured from the onset of the target/terminal word-stream stimuli.
!
19
Participants were instructed to ignore the continuous stream of words in one channel and
to pay attention to the pairs of words in the other channel, and to indicate whether the second
item in each attended pair was a real word in English or not by pressing the green button (yes) or
the red button (no) on the response box. The order of the buttons on the response box (i.e., green
on the left or right) was counterbalanced across participants. Participants were instructed to use
the index finger of their right (dominant) hand to make a response, to rest their finger between
the two buttons after each trial, and to respond as quickly and as accurately as possible. A series
of 12 practice trials was presented prior to the start of the experiment to familiarize participants
with the task.
3. Participants
Participants were recruited from Birkbeck College and the local community via the Sona
Experiment Management System and fliers posted on campus. Sample size was determined a
priori based on the number of participants per experimental condition in Experiment 1 (N=8 per
priming condition), and no additional participants were tested after these data had been collected.
Thirty-two right-handed native speakers of British English (13 male, 19 female) aged 18 to 45
years (mean = 28.38, SD = 8.33) with no reported hearing impairment or neurological illness
took part in the experiment in exchange for payment or course credit.!Participants recruited via
Sona were selected for participation according to their declared handedness (left-handed
participants were automatically excluded). In addition, all participants were given a
questionnaire immediately prior to testing in which they were asked to confirm that they were
right-handed and had normal hearing.
!
20
B. Results
Accuracy (percent correct) and harmonic mean RTs for correct responses were calculated
for each participant in each condition. Average values in each condition are shown in Table II.
Repeated measures ANOVAs were conducted on the accuracy and RT data, with Attended Prime
(related or unrelated) and Competing Prime (related or unrelated) as within-subjects variables
and Ear of Presentation (right or left) as a between subjects variable. The results of the accuracy
analysis revealed a significant main effect of Attended Prime, F(1,30) = 44.88, p < .0001, partial
η2 = .599, such that participants were more accurate in response to targets preceded by a related
prime (mean % correct = 98.23, SD = 2.24) than an unrelated prime (mean % correct = 92.50,
SD = 5.15). There were no significant main effects of Competing Prime, F(1,30) = 1.46, p = .237,
partial η2 = .046, or Ear of Presentation, F(1,30) = 0.43, p = .515, partial η2 = .014, and no
significant interactions, all F < 2.51, all p > .124, all partial η2 < .077.
Similar results were obtained in the RT analysis: a significant main effect of Attended
Prime emerged, F(1,30) = 31.58, p < .0001, partial η2 = .513, with faster RTs obtained to targets
preceded by related primes than unrelated primes (see Table III). There was no significant main
effect of Competing Prime, F(1,30) = 0.22, p = .646, partial η2 = .007. The magnitude of the
semantic priming effect for Attended and Competing Primes is shown in Table III. Thus, in the
RT data, participants showed significant semantic priming for attended primes, but not for
competing primes. The main effect of Ear of Presentation was not significant, F(1,30) = 2.15, p
= .153, partial η2 = .067, and there were no significant interactions, all F < 1.29, all p > .265, all
partial η2 < .041.
!
21
C. Discussion
As predicted, a robust semantic priming effect emerged for prime words presented in the
attended channel. However, semantic priming was not obtained for primes presented
simultaneously in the unattended channel. This result is inconsistent with the proposal that
unattended speech is processed at the semantic level in dichotic listening under the conditions in
this experiment, i.e., attended speech with a F0 of +40 Hz relative to the unattended speech
signal presented at a SNR of +12 dB. These findings extend previous work by Rivenez et al.
(2006), who observed identity priming of attended targets by unattended primes under similar
conditions. Rivenez et al. (2006) propose that identity priming effects emerge for unattended
speech only when the attended and competing signals are differentiated by acoustic cues that
allow for perceptual grouping of the two speech streams, such as F0, as these cues serve to
enhance perceptual clarity when the unattended signal is attenuated by 12 dB relative to the
attended signal. The absence of a semantic priming effect for unattended words in the present
study suggests that the results obtained by Rivenez et al. (2006) were due to the shared lexical
form of the prime and target, and that the processing of unattended speech stimuli under these
conditions is limited to form-based analysis.
Alternatively, the form-based priming observed in Rivenez et al. (2006) may be
attributable to the presentation of unattended primes in the context of a sentence, as opposed to a
continuous series of words as in the present study. As unattended primes in Rivenez et al. (2006)
were presented as final words in naturally produced sentences, the ongoing acoustic context
provided both prosodic and syntactic cues to the location of the prime. This may have enhanced
the salience of the unattended primes relative to the present study, in which unattended primes
were presented in a continuous series of words produced by the same talker with no acoustic or
!
22
linguistic cues to differentiate the prime from other words in the series. Although Rivenez et al.
(2006) used a secondary task to ensure that listeners’ attention was focused away from
unattended primes in their study, further research is needed to establish whether form-based
priming would be observed under the conditions used in the present experiment. Nevertheless,
the present findings indicate that a 40 Hz F0 difference between attended and competing stimuli
did not facilitate semantic processing of unattended speech at this SNR.
Neither the present study nor Rivenez et al. (2006) found evidence of ear-of-presentation
effects on responses to attended targets. This result is consistent with previous findings
demonstrating that the REA for speech stimuli in dichotic listening depends upon the relative
intensity of the two signals. When the interaural intensity difference is +12 dB in favour of left-
ear stimuli, as in the present study, a left-ear advantage (LEA) is observed such that the overall
laterality effect is eliminated (Hugdahl et al., 2008; Westerhausen et al., 2009).
Although the present experiment did not find evidence of semantic processing of
unattended speech, previous studies have shown that higher-level processing of an unattended
signal depends critically upon the acoustic characteristics of the attended and competing speech
stimuli (Rivenez et al., 2006, 2007, 2008). The present findings suggest that perceptual grouping
based on F0 differences is not sufficient to allow for the activation of semantic information by an
unattended speech signal presented dichotically at a relative intensity level of –12 dB. However,
semantic processing of unattended speech may occur when the relative intensity of the
competing signal is increased. The meaningful content of a competing speech signal has been
shown in previous studies to affect the comprehension of an attended sentence in dichotic
listening at a SNR of 0 dB (Aydelott et al., 2012), which reflects the listening demands of a
typical cocktail party (Plomp, 1977; cf. Bronkhorst, 2000). This suggests that semantic
!
23
information from an unattended signal can influence the processing of an attended signal under
more demanding listening conditions. Experiment 3 tested this hypothesis by examining whether
semantic priming of attended targets by unattended primes would emerge when the attended and
competing signals were presented at the same intensity level.
IV. EXPERIMENT 3
A. Method
1. Stimuli
The stimuli were identical to those in Experiment 2, with the exception that all stimuli
(attended and competing primes, targets, and distractor words) were scaled to the same nominal
mean rms amplitude in Praat, yielding a SNR of 0 dB, i.e., the attended stimuli (signal) and the
competing stimuli (noise) were presented at the same intensity level.
2. Procedure
The procedure, equipment, and presentation level were identical to those in Experiment 2.
3. Participants
Participants were recruited from Birkbeck College and the local community via the Sona
Experiment Management System. Sample size was determined a priori based on the number of
participants in Experiment 2, and no additional participants were tested after these data had been
collected. Thirty-two right-handed native speakers of British English (12 male, 20 female) aged
18 to 45 years (mean = 29.13, SD = 7.95) with no reported hearing impairment or neurological
illness took part in the experiment in exchange for payment or course credit. As in Experiment 2,
participants recruited via Sona were selected for participation according to their declared
handedness (left-handed participants were automatically excluded), and all participants were
!
24
given a questionnaire immediately prior to testing in which they were asked to confirm that they
were right-handed and had normal hearing. None had participated in Experiment 2.
B. Results
Accuracy (percent correct) and harmonic mean RTs for correct responses were calculated
for each participant in each condition. Average values in each condition are shown in Table IV.
Repeated measures ANOVAs were conducted on the accuracy and RT data, with Attended Prime
(related or unrelated) and Competing Prime (related or unrelated) as within-subjects variables
and Ear of Presentation (right or left) as a between subjects variable. The results of the accuracy
analysis revealed a significant main effect of Attended Prime, F(1,30) = 17.14, p < .0001, partial
η2 = .364, such that participants were more accurate in responding to targets preceded by a
related prime (mean % correct = 95.21, SD = 4.93) than an unrelated prime (mean % correct =
90.00, SD = 8.21). There were no significant main effects of Competing Prime, F(1,30) = 2.12, p
= .156, partial η2 = .066, or Ear of Presentation, F(1,30) = 0.84, p = .367, partial η2 = .027, and
no significant interactions, all F < 0.69, all p > .414, all partial η2 < .022.
The results of the RT analysis revealed significant effects of both Attended Prime,
F(1,30) = 40.90, p < .0001, partial η2 = .577, and Competing Prime, F(1,30) = 5.46, p = .026
partial η2 = .154. Thus, unlike in Experiment 2, both attended and competing prime stimuli
produced significant semantic priming. The magnitude of the semantic priming effect for
Attended and Competing Primes is shown in Table V. As the design of Experiment 3 was
identical to that of Experiment 2, it was possible to compare effect sizes directly using partial η2
(Fritz et al., 2012). Partial η2 provides an appropriate summary of effect size for phenomena such
as semantic priming that are necessarily examined using a within-subjects design (Lakens, 2013).
!
25
Partial η2 values for the main effect of Competing Prime in Experiments 2 and 3 are shown in
Tables III and V, respectively. At a +12 dB SNR (Experiment 2), competing primes had only a
very small effect on RTs to targets (partial η2 < .01), whereas a larger effect (partial η2 > .15) was
observed at a 0 dB SNR (Experiment 3). The main effect of Ear of Presentation was at the
threshold of significance, F(1,30) = 3.96, p = .056, partial η2 = .117, such that RTs were faster to
targets presented to the right ear (mean = 1281, SD = 131) than the left ear (mean = 1367, SD =
113). There were no significant interactions, all F < 0.58, all p > .451, all partial η2 < .019.
C. Discussion
As in Experiment 2, attended primes facilitated responses to semantically related targets
in the attended channel in Experiment 3. However, in contrast to the results of Experiment 2, a
significant semantic priming effect was also obtained in response to unattended primes in the RT
data, indicating that unattended stimuli were processed at the semantic level in Experiment 3.
Thus, an increase in the relative intensity of the competing signal from +12 dB to 0 dB SNR was
sufficient to allow semantic activation by unattended speech presented at a F0 of –40 Hz. In
addition, unlike in Experiment 2, the ear of presentation of the attended stimuli influenced RTs
to target words, such that lexical decision judgments were faster in response to targets presented
to the right ear than the left ear. Thus, whereas a REA for speech stimuli was not observed at a
SNR of +12 dB, targets presented to the right channel were processed more efficiently than
targets presented to the left channel at a SNR of 0 dB.
!
26
V. GENERAL DISCUSSION
This study investigated the conditions under which semantic information from an
unattended speech signal may influence the processing of attended speech in dichotic listening.
A lexical decision paradigm was administered in which target words presented in the attended
channel were immediately preceded by two different prime words presented simultaneously, one
in the attended channel and the other in the unattended channel. Both the attended and competing
primes were either semantically related or unrelated to the target word, and semantic priming
(faster RTs to attended targets preceded by related primes than by unrelated primes) served as an
index of semantic processing. The F0 of the attended stimuli was increased by 40 Hz relative to
the unattended stimuli to facilitate perceptual grouping (cf. Rivenez et al., 2006). To discourage
attentional switches to the competing prime stimuli, the attended prime–target pairs were
presented in isolation, whereas competing primes in the unattended channel were presented in a
continuous stream of words (cf. Dupoux et al., 2003).
The results revealed that, when the unattended signal was attenuated by 12 dB relative to
the attended signal, semantic priming was obtained only for attended primes and not for
competing primes. However, when the competing signal was presented at a SNR of 0 dB,
semantic priming also emerged for competing primes. Thus, semantic processing of unattended
speech stimuli may be observed in dichotic listening for attended and competing speech signals
that differ in pitch when the two streams are presented at the same intensity level. This result
confirms and extends previous findings demonstrating that higher-level processing of unattended
speech is highly sensitive to the acoustic properties of the attended and competing signals.
Previous studies have indicated that lexical processing of unattended stimuli in dichotic listening
at a SNR of +12 dB depends upon the segregation of the attended and competing streams based
!
27
on talker characteristics such as F0, timbre, and vocal tract length (Rivenez et al., 2006, 2007,
2008). Whether the lexical priming effects observed in these previous studies are partially
influenced by the presentation of unattended stimuli in a sentence context, as opposed to a
continuous stream of words as in the present study, requires further investigation. However, the
present findings suggest that the relative intensity of the unattended signal must also be increased
before access to semantic information is achieved.
These results also suggest that the meaningful content of unattended speech may be more
likely to intrude on the processing of an attended message when the competing signal is
presented at the same intensity level as the attended signal. This is consistent with previous
findings from the classic dichotic listening paradigm, in which the attended and competing
signals are typically matched for loudness, and intrusions from the unattended channel are
observed when the meaning of an unattended stimulus is salient or relevant to the attended
message (e.g., Moray, 1959; Treisman, 1960; Mackay, 1973; Holender, 1986; cf. Bentin et al.,
1995). However, the extent to which these earlier results reflect semantic activation in the
absence of attention, as opposed to attentional switches leading to conscious identification of
stimuli in the unattended channel, has since been called into question (Holender, 1986; Dupoux
et al., 2003; Lachter, Forster, & Ruthruff, 2004; but cf. Grainger & Holcomb, 2014). The present
study was designed to minimize shifts of attention by presenting the attended prime–target pairs
in isolation while the unattended primes were presented in continuous speech, thereby enhancing
the salience of the attended stimuli (cf. Dupoux et al., 2003). This method proved effective in
Experiment 2: unattended primes failed to influence the processing of attended targets at a SNR
of +12 dB, even though the unattended primes were fully realized words that produced semantic
priming at the same relative intensity level in Experiment 1 when presented without continuous
!
28
speech under conditions of divided attention. Nevertheless, it is possible that attentional switches
contributed to the semantic priming effect observed at the 0 dB SNR in Experiment 3. An
increase in the relative intensity of the competing speech stimuli may have made them more
salient to listeners, and therefore more likely to capture attention away from the attended channel.
Further, as stimulus intensity differences can aid perceptual selection in dichotic listening (cf.
Bloch & Hellige, 1989), the presentation of the two signals at a 0 dB SNR may have increased
demands on auditory segregation, resulting in greater interference from the competing signal.
Future research using a secondary measure to control for attentional switches, as in Rivenez et al.
(2006), may serve to establish whether these results can be attributed to semantic activation by
unattended speech without conscious identification.
The relative intensities of the attended and competing stimuli also contributed to the
emergence of ear-of-presentation effects in the present study. Consistent with the findings of
Rivenez et al. (2006), no effect of attended ear was observed at a +12 dB SNR in Experiment 2.
Thus, the pattern of results did not differ as a function of left or right ear of presentation at this
SNR, suggesting that attention to the relevant channel served to override the REA for speech
stimuli observed in previous dichotic listening studies (Kimura, 1961, 1967; cf. Hugdahl et al.,
2009). However, the results of Experiment 3 revealed an effect of attended ear consistent with
the REA at a SNR of 0 dB, such that RTs were faster to target stimuli presented in the right
channel. The contribution of stimulus intensity differences to ear-of-presentation effects in
dichotic listening has been examined extensively by Hugdahl and colleagues, who report that the
attenuation of a right-ear stimulus by –9 to –15 dB relative to a left-ear stimulus is sufficient to
reverse the REA in favour of the left channel, even in the absence of selective attention (Hugdahl
et al., 2008; Westerhausen et al., 2009). In contrast, at a SNR of 0 dB, selective attention to the
!
29
left channel is necessary to counteract the REA, whereas attention to the right channel increases
the magnitude of the advantage (Westerhausen et al., 2009). Thus, the present findings can be
accounted for in terms of the interaction of SNR and attention in dichotic listening: when
attended targets are presented at a greater intensity level than competing words, speech
identification processes will tend to favour the attended stimuli irrespective of ear of presentation,
but when attended and competing stimuli are presented at the same intensity level, the enhanced
REA for attended stimuli presented in the right channel may result in faster processing of targets
in the right-ear condition.
It is worth noting, however, that the REA observed at the 0 dB SNR did not extend to the
semantic processing of the prime stimuli. No significant interaction of ear of presentation and
priming was observed for attended or competing primes in Experiment 3; thus, there was no
difference in the magnitude of the semantic priming effect produced by attended or unattended
primes due to the ear of presentation of the stimuli. This suggests that the auditory channel in
which the primes were presented did not influence the semantic activation produced by the prime
stimuli or its implicit effect on related targets. Instead, the REA was apparent only in explicit
lexical decision responses to target words. The lexical decision task required listeners to isolate
the target from the competing signal and analyze its phonological form, similar to the speech
identification tasks used in previous dichotic listening studies (Hugdahl et al., 2008;
Westerhausen et al., 2009). Explicit judgments requiring selective phonological analysis of one
of two competing stimuli may place greater demands on speech processing than implicit priming,
and may therefore be more sensitive to factors such as the REA that facilitate the identification
of speech stimuli.
!
30
The present findings have potential implications for the study of language comprehension
in a multitalker environment. Identifying the conditions under which listeners are sensitive to
semantic information in unattended speech may provide further insight into the disruption of
higher-level language processing in complex listening situations. Previous research has shown
that multitalker babble consisting of natural speech is a more effective masker than babble-
modulated noise, which has been attributed to “linguistic” interference by the natural speech
stimuli (Simpson & Cooke, 2005; cf. Van Engen & Bradlow, 2007). Subsequent studies have
demonstrated that the linguistic content of a competing signal contributes to its masking effect,
particularly when the number of competing talkers is small: two-talker babble is more disruptive
to speech recognition when the interfering signal is in the listener’s native language as opposed
to a foreign language (Van Engen & Bradlow, 2007; cf. Garcia Lecumberri & Cooke, 2006) or
consists of high-frequency words as opposed to low-frequency words (Boulenger et al., 2010).
These findings indicate that a meaningful unattended speech signal introduces irrelevant
linguistic information that may interfere with the comprehension of the attended message.
However, these studies have not addressed whether this interference is due to the semantic
content of the competing signal, as opposed to other linguistic information in the unattended
message, e.g., phonological or lexical cues (cf. Van Engen & Bradlow, 2007).
In contrast, the present study used a direct measure of semantic activation to determine
whether unattended speech is processed at the semantic level. The results of Experiment 3
suggest that speech produced by a single competing talker at a different pitch and spatial location
relative to the attended talker produces significant semantic activation when the two signals are
presented at the same intensity level. The meaningful content of a competing speech signal is
therefore likely to affect the higher-level comprehension of attended speech under these
!
31
conditions. This is consistent with the findings of Aydelott et al. (2012), who observed that a
meaningful competing speech signal presented at a SNR of 0 dB reduced access to sentence-
level semantic cues in attended speech in dichotic listening, whereas a time-reversed version of
the same signal did not affect the processing of sentence-level meaning. However, improvements
in SNR appear to decrease the likelihood of intrusions of semantic information from an irrelevant
speech stream, as indicated by the findings of Experiment 2: when the competing speech signal
was attenuated by 12 dB relative to the attended signal, no evidence of semantic processing of
unattended speech was observed. Thus, under these conditions, selective attention to the attended
channel was sufficient to prevent semantic activation by unattended speech. The extent to which
listeners are able to attend selectively to one signal without the intrusion of semantic content
from another is likely to be an important factor in the successful interpretation of spoken
language in the presence of multiple simultaneous talkers.
In summary, the results of the present study demonstrate that semantic processing of
unattended speech in dichotic listening depends critically upon the acoustic properties of the
attended and competing signals. When the attended and competing speech streams can be
differentiated on the basis of F0, the relative intensity of the two signals determines whether the
unattended signal is processed at the semantic level. Semantic information from the irrelevant
channel does not influence responses to attended stimuli when the unattended signal is attenuated
by 12 dB, indicating that listeners are able to ignore the meaningful content of the competing
message under these conditions. In contrast, semantic priming of attended targets by unattended
primes is observed when the attended and competing signals are presented at a SNR of 0 dB,
indicating that selective attention fails to inhibit semantic processing of unattended speech
presented at the same intensity level as the attended signal. Thus, an increase in the relative
!
32
intensity of the competing signal is sufficient to result in semantic activation by unattended
speech. Whether this reflects the increased intelligibility of the competing signal, or an increased
likelihood of attentional switches to the unattended channel, remains a topic for future research.
!
33
ACKNOWLEDGMENTS
The authors would like to thank three anonymous reviewers for their comments on a previous
version of this manuscript. The authors would also like to thank Sarah Cunningham for her
assistance in recording and editing a subset of the stimuli used in these experiments; Jonathan
Reynolds and Fiona Reynolds for conducting preliminary experiments; and Angelina Koleva and
Lauren Williams for collecting preliminary data. This research was supported by the Department
of Psychological Sciences, Birkbeck, University of London.
!
34
APPENDIX
Related Primes Unrelated Primes Targets
Attended Competing Attended Competing
queen throne bread swamp king
june march oil bed month
pub lounge smog duck bar
meal eat van lie food
throw catch smoke shift ball
cure spa pot broom health
ride cart choose sock horse
swim dive trick bid pool
span forth rob dust bridge
weigh gauge ache leaf scale
page read snow hope book
throb pulse mug rack heart
lock clasp sheep fret chain
pass fail camp glance test
ring call suit peace phone
palm glove joke spy hand
robe gown nail kick dress
mop sweep leash pill floor
time place say help date
chore task lamb block job
gasp air hem street breath
sleep trance fund tug dream
stool bench glue fruit chair
talk voice lot age speech
day hour fact plant week
sprint track harp mood race
choke cough pearl weed throat
spouse bride elm tape wife
pad write jig reach note
form size church ask shape
pal mate leak grip friend
ear chin fence cloud nose
look gaze set boil view
sun earth chief hall moon
loft beam fade cope roof
grave tomb rice hub death
crowd bunch lake flame group
ship yacht dance gem boat
gram ton mouse cake weight
learn class touch hair school
!
35
arm thigh source dusk leg
cub wolf prank spoon bear
smile frown shop kite face
clerk pen sing flood desk
head scarf war raft neck
beer drink phrase post wine
room house give think wall
knob hinge sip crease door
tree grass fat spend park
thorn bud hoop chalk rose
lip tongue bell yard mouth
bus car fish word train
ray spark twist egg light
court trial land step judge
team score knife fig goal
case sue night tap law
wing flight kiss salt plane
cheap save fog taste deal
reel loop wade cat film
sand shore coal gain beach
!
36
REFERENCES
Aydelott, J., Baer-Henney, D., Trzaskowski, M., Leech, R. & Dick, F. (2012). Sentence
comprehension in competing speech: Dichotic sentence-word priming reveals hemispheric
differences in auditory semantic processing. Language and Cognitive Processes, 27, 1108–1134.
Bentin, S., Kutas, M., & Hillyard, S. A. (1995). Semantic processing and memory for
attended and unattended words in dichotic listening: Behavioral and electrophysiological
evidence. Journal of Experimental Psychology: Human Perception and Performance, 21(1), 54.
Bloch, M. I., & Hellige, J. B. (1989). Stimulus intensity, attentional instructions, and the
ear advantage during dichotic listening. Brain and Cognition, 9(1), 136–148.
Boersma, P. & Weenink, D. (2010). Praat: Doing phonetics by computer [Computer
program]. Version 5.1.45, retrieved from http://www.praat.org/
Boulenger, V., Hoen, M., Ferragne, E., Pellegrino, F., & Meunier, F. (2010). Real-time
lexical competitions during speech-in-speech comprehension. Speech Communication, 52(3),
246–253.
Broadbent, D. E. (1958). Perception and Communication. London, UK: Pergamon Press.
Bronkhorst, A. W. (2000). The cocktail party phenomenon: A review of research on
speech intelligibility in multiple-talker conditions. Acta Acustica, 86, 117–128.
Cherry, E. C. (1953). Some experiments on the recognition of speech, with one and two
ears. The Journal of the Acoustical Society of America, 25(5), 975–979.
Coltheart, M. (1981). The MRC Psycholinguistic Database. Quarterly Journal of
Experimental Psychology, 33A, 495–505.
!
37
Dupoux, E., Kouider, S., & Mehler, J. (2003). Lexical access without attention?
Explorations using dichotic priming. Journal of Experimental Psychology: Human Perception
and Performance, 29(1), 172–183.
Fritz, C. O., Morris, P. E., & Richler, J. J. (2012). Effect size estimates: Current use,
calculations, and interpretation. Journal of Experimental Psychology: General, 141(1), 2–18.
Garcia Lecumberri, M. L., & Cooke, M. (2006). Effect of masker type on native and
non-native consonant perception in noise. The Journal of the Acoustical Society of America,
119(4), 2445–2454.
Grainger, J., & Holcomb, P. (2014). An ERP investigation of dichotic repetition priming
with temporally overlapping stimuli. Psychonomic Bulletin & Review. doi:10.3758/s13423-014-
0677-3
Gray, J. A., & Wedderburn, A. A. I. (1960). Grouping strategies with simultaneous
stimuli. Quarterly Journal of Experimental Psychology, 12, 180–184.
Holender, D. (1986). Semantic activation without conscious identification in dichotic
listening, parafoveal vision, and visual masking: A survey and appraisal. The Behavioral and
Brain Sciences, 9, 1–66.
Hugdahl, K. (2005). Symmetry and asymmetry in the human brain. European Review,
13(S2), 119–133.
Hugdahl, K., Westerhausen, R., Alho, K., Medvedev, S. & Hämäläinen, H. (2008). The
effect of stimulus intensity on the right ear advantage in dichotic listening. Neuroscience Letters,
431, 90–94.
!
38
Hugdahl, K., Westerhausen, R., Alho, K., Medvedev, S., Laine, M., & Hämäläinen, H.
(2009). Attention and cognitive control: unfolding the dichotic listening story. Scandinavian
Journal of Psychology, 50(1), 11–22.
Kimura, D. (1961). Cerebral dominance and the perception of verbal stimuli. Canadian
Journal of Psychology, 15(3), 166–171.
Kimura, D. (1967). Functional asymmetry of the brain in dichotic listening. Cortex, 3,
163–178.
Kiss, G. R., Armstrong, C., Milroy, R., & Piper, J. (1973) An associative thesaurus of
English and its computer analysis. In A. J. Aitken, R. W. Bailey, & N. Hamilton-Smith, (Eds.),
The Computer and Literary Studies (pp. 153–165). Edinburgh, UK: Edinburgh University Press.
Lachter, J., Forster, K. I., & Ruthruff, E. (2004). Forty-five years after Broadbent (1958):
still no identification without attention. Psychological Review, 111(4), 880–913.
Lakens, D. (2013). Calculating and reporting effect sizes to facilitate cumulative science:
A practical primer for t-tests and ANOVAs. Frontiers in Psychology, 4(863).
doi:10.3389/fpsyg.2013.00863
Mackay, D. G. (1973) Aspects of the theory of comprehension, memory and attention.
Quarterly Journal of Experimental Psychology, 25, 22–40.
McNamara, T. P. (2005). Methodological issues. In Semantic priming: Perspectives from
memory and word recognition (pp. 44–54). Hove, UK: Psychology Press.
Moray, N. (1959). Attention in dichotic listening: Affective cues and the influence of
instructions. The Quarterly Journal of Experimental Psychology, 11(1), 56–60.
Plomp, R. (1977). Acoustical aspects of cocktail parties. Acustica, 38, 186–191.
!
39
Ratcliff, R. (1993). Methods for dealing with reaction time outliers. Psychological
Bulletin, 114(3), 510–532.
Rivenez, M., Darwin, C. J., & Guillaume, A. (2006). Processing unattended speech.
Journal of the Acoustical Society of America, 119, 4027–4040.
Rivenez, M., Darwin, C. J., Bourgeon, L., & Guillaume, A. (2007). Unattended speech
processing: Effect of vocal tract length. Journal of the Acoustical Society of America, 121,
EL90–EL95.
Rivenez, M., Guillaume, A., Bourgeon, L., & Darwin, C. J. (2008). Effect of voice
characteristics on the attended and unattended processing of two concurrent messages. European
Journal of Cognitive Psychology, 20(6), 967–993.
Schneider, B. A., Li, L. & Daneman, M. (2007). How competing speech interferes with
speech comprehension in everyday listening situations. Journal of the American Academy of
Audiology, 18, 559–572.
Simpson, S. A., & Cooke, M. (2005). Consonant identification in N-talker babble is a
nonmonotonic function of N. Journal of the Acoustical Society of America, 118(5), 2775–2778.
Treisman, A. M. (1960) Contextual cues in selective listening. Quarterly Journal of
Experimental Psychology, 12, 242–248.
Treisman, A. M., & Riley, J. G. A. (1969) Is selective attention selective perception or
selective response? A further test. Journal of Experimental Psychology, 79, 27–34.
Van Engen, K. J., & Bradlow, A. R. (2007). Sentence recognition in native- and foreign-
language multi-talker background noise. The Journal of the Acoustical Society of America,
121(1), 519–526.
!
40
Van Heuven, W. J. B., Mandera, P., Keuleers, E., & Brysbaert, M. (2014). SUBTLEX-
UK: A new and improved word frequency database for British English. Quarterly Journal of
Experimental Psychology. doi:10.1080/17470218.2013.850521
Westerhausen, R., Moosmann, M., Alho, K., Medvedev, S., Hämäläinen, H., & Hugdahl,
K. (2009). Top–down and bottom–up interaction: manipulating the dichotic listening ear
advantage. Brain Research, 1250, 183–189.
Wood, N. L., & Cowan, N. (1995a). The cocktail party phenomenon revisited: Attention
and memory in the classic selective listening procedure of Cherry (1953). Journal of
Experimental Psychology: General, 124(3), 243–262.
Wood, N. L., & Cowan, N. (1995b). The cocktail party phenomenon revisited: How
frequent are attention shifts to one's name in an irrelevant auditory channel? Journal of
Experimental Psychology: Learning, Memory, and Cognition, 21, 255–260.
!
41
TABLE I
Mean (and standard deviation) frequency of occurrence, length, and word association data for
attended and competing primes used in Experiments 1, 2, and 3.
Semantic
relationship to
target
Prime condition
SUBTLEX-UK
Zipf frequency
value
Length in
phonemes
Related
Attended
4.43 (0.83)
3.21 (0.74)
Competing
4.35 (0.69)
3.30 (0.65)
Unrelated
Attended
4.50 (0.72)
3.28 (0.67)
Competing
4.45 (0.68)
3.47 (0.75)
!
42
TABLE II
Results of Experiment 2 (+12 dB SNR). Mean (and standard deviation) RT and accuracy of
responses to targets preceded by attended and competing primes as a function of semantic
relationship between primes and targets.
Attended prime
Competing prime
Ear of
presentation
RT (msec)
Related
Related
Left
1381 (199)
Right
1302 (134)
Unrelated
Left
1394 (230)
Right
1270 (88)
Unrelated
Related
Left
1443 (179)
Right
1380 (165)
Unrelated
Left
1470 (233)
Right
1399 (146)
!
43
TABLE III
Magnitude of the semantic priming effect produced by attended and competing primes in
Experiment 2 (+12 dB SNR).
Prime
condition
Semantic
condition
Mean (SD)
RT
Semantic
priming
p-value
Effect size
(partial η2)
Attended
Related
1337 (171)
86
<.0001
.513
Unrelated
1423 (172)
Competing
Related
1377 (162)
7
.646
.007
Unrelated
1383 (178)
!
44
TABLE IV
Results of Experiment 3 (0 dB SNR). Mean (and standard deviation) RT and accuracy of
responses to targets preceded by attended and competing primes as a function of semantic
relationship between primes and targets.
Attended prime
Competing prime
Ear of
presentation
RT (msec)
Related
Related
Left
1296 (88)
Right
1232 (152)
Unrelated
Left
1337 (155)
Right
1235 (110)
Unrelated
Related
Left
1408 (170)
Right
1315 (144)
Unrelated
Left
1425 (135)
Right
1341 (162)
!
45
TABLE V
Magnitude of the semantic priming effect produced by attended and competing primes in
Experiment 3 (0 dB SNR).
Prime
condition
Semantic
condition
Mean (SD)
RT
Semantic
priming
p-value
Effect size
(partial η2)
Attended
Related
1275 (125)
97
<.0001
.577
Unrelated
1372 (144)
Competing
Related
1313 (130)
22
.026
.154
Unrelated
1334 (131)
!
46
FIGURE 1
attended
prime
competing
prime
target
distractor
word
distractor
word
distractor
word
distractor
word
distractor
word
distractor
word
distractor
word
ATTENDED CHANNEL
(+40 Hz)
LEAD-IN WORD STREAM
PRIME
TARGET + TERMINAL WORD STREAM
UNATTENDED CHANNEL
!
47
FIGURE CAPTION
Figure 1
Illustration of the structure of experimental trials in Experiments 2 and 3. This schematic
represents the sequence of events in a single trial. Distractor words and competing prime stimuli
were presented with an inter-stimulus interval of 0 msec, as were attended prime–target pairs,
and trials were presented with an inter-trial interval of 0 msec. Thus, stimuli in the unattended
channel were presented as a continuous stream of words, and stimuli in the attended channel
were presented as intermittent prime–target pairs.
... distractors (Rhebergen et al. 2005;Iyer et al. 2010;Best et al. 2012;Gallun and Diedesch 2013;72 Carey et al. 2014;Kilman et al. 2014;Swaminathan et al. 2015;Kidd et al. 2016), and there are 73 also indications for implicit processing of the semantic content of task-irrelevant speech, 74 manifest through priming effects or memory intrusions (Tun et al. 2002;Dupoux et al. 2003;75 Rivenez et al. 2006;Beaman et al. 2007;Carey et al. 2014;Aydelott et al. 2015;Schepman et al. 76 attributes that contribute to energetic masking (e.g. loudness, pitch, and fine-structure), as well 111 ...
... Sound Effect' (ISE), showing that the semantic content of task-irrelevant input affects 542 performance on a main task, mainly through priming effects and interference with short-term 543 memory (Lewis 1970;Bentin et al. 1995;Surprenant et al. 1999;Dupoux et al. 2003;Beaman 544 2004;Rivenez et al. 2006;Beaman et al. 2007;Rämä et al. 2012;Aydelott et al. 2015;Schepman 545 et al. 2016;Vachon et al. 2019). However, an important caveat precludes interpreting these 546 findings as clear-cut evidence for semantic processing of task-irrelevant speech: Since these 547 studies primarily involve presentation of arbitrary lists of words (mostly nouns), usually at a 548 relatively slow rate, an alternative explanation is that the ISE is simply a result of occasional 549 shifts of attention towards task-irrelevant stimuli (Carlyon 2004;Lachter et al. 2004). ...
... have the advantage of probing behavior at a finer scale, but are substantially less ecological 744 (e.g., memory-recall for short lists of words or priming effects; (Tun et al. 2002;Dupoux et al. 745 2003;Rivenez et al. 2006Rivenez et al. , 2008Carey et al. 2014;Aydelott et al. 2015). In moving towards 746 studying speech processing and attention under more ecological circumstances, using natural 747 continuous speech, we face an experimental challenge of obtaining sufficiently sensitive 748 behavior measures without disrupting listening with an ongoing task (e.g., target detection) or 749 ...
Article
Full-text available
Paying attention to one speaker in noisy environments can be extremely difficult, because to-be-attended and task-irrelevant speech compete for processing resources. We tested whether this competition is restricted to acoustic-phonetic interference or if it extends to competition for linguistic processing as well. Neural activity was recorded using Magnetoencephalography as human participants were instructed to attended to natural speech presented to one ear, and task-irrelevant stimuli were presented to the other. Task-irrelevant stimuli consisted either of random sequences of syllables, or syllables structured to form coherent sentences, using hierarchical frequency-tagging. We find that the phrasal structure of structured task-irrelevant stimuli was represented in the neural response in left inferior frontal and posterior parietal regions, indicating that selective attention does not fully eliminate linguistic processing of task-irrelevant speech. Additionally, neural tracking of to-be-attended speech in left inferior frontal regions was enhanced when competing with structured task-irrelevant stimuli, suggesting inherent competition between them for linguistic processing.
... By allocating cognitive resources to a parallel task, recognition scores can be brought below ceiling such that turn-taking effects are revealed for alternating talkers (Schneider et al., 2007). This approach is based on limited cognitive resources and assumes that unattended speech is semantically processed (Aydelott et al., 2015;Bentin et al., 1995). An example of a one-talker-at-a-time test with high ecological validity is the National Acoustic Laboratories Dynamic Conversations Test (NAL-DCT, Best et al., 2016Best et al., , 2018. ...
... This is consistent with Best et al. (2008), who found that switching costs increase with increasing follow-up speed of sentences, due to interruptions in the continuity of auditory objects. Moreover, more speech per time increases cognitive load (Müller et al., 2019), even if only unattended speech is added (Aydelott et al., 2015;Bentin et al., 1995). ...
Article
Full-text available
A multi-talker paradigm is introduced that uses different attentional processes to adjust speech-recognition scores with the goal of conducting measurements at high signal-to-noise ratios (SNR). The basic idea is to simulate a group conversation with three talkers. Talkers alternately speak sentences of the German matrix test OLSA. Each time a sentence begins with the name “Kerstin” (call sign), the participant is addressed and instructed to repeat the last words of all sentences from that talker, until another talker begins a sentence with “Kerstin”. The alternation of the talkers is implemented with an adjustable overlap time that causes an overlap between the call sign “Kerstin” and the target words to be repeated. Thus, the two tasks of detecting “Kerstin” and repeating target words are to be done at the same time. The paradigm was tested with 22 young normal-hearing participants (YNH) for three overlap times (0.6 s, 0.8 s, 1.0 s). Results for these overlap times show significant differences, with median target word recognition scores of 88%, 82%, and 77%, respectively (including call-sign and dual-task effects). A comparison of the dual task with the corresponding single tasks suggests that the observed effects reflect an increased cognitive load.
... Consequently, when listening to speech-in-speech with selective attention, the recognition of attended speech is increased while unattended speech is ignored to some degree. Still, unattended speech is semantically processed (Aydelott et al., 2015;Bentin et al., 1995) and thus influences recognition of attended speech by blocking resources (Schneider et al., 2007). Recognizing simultaneous speech with divided attention is also possible, but it is limited to a few words (for two talkers: 2-3 words each) and depends on the background noise (Best et al., 2010;Meister et al., 2018). ...
... Considering that also unattended speech is semantically processed (Aydelott et al., 2015;Bentin et al., 1995), this may be relevant for both target and concurrent speech. ...
Preprint
Full-text available
A multi-talker paradigm is introduced that uses different attentional processes to adjust speech recognition scores with the goal to conduct measurements at high signal-to-noise ratios. The basic idea is to simulate a group conversation with three talkers and a participant. Talkers alternately speak sentences of the German matrix test OLSA. Each time a sentence begins with the name "Kerstin" (call sign), the participant is addressed and instructed to repeat the last words of all sentences from that talker, until another talker begins a sentence with "Kerstin". The alternation of the talkers is implemented with an adjustable overlap time that causes an overlap between the call sign "Kerstin" and the target words to be repeated. Thus, the two tasks of detecting "Kerstin" and repeating target words are to be processed at the same time as a dual task. The paradigm was tested with 22 young normal-hearing participants for three overlap times (0.6 s, 0.8 s, 1.0 s). Results for these overlap times show significant differences with median target word recognition scores of 88%, 82%, and 77%, respectively (including call sign and dual task effects). A comparison of the dual task with the corresponding single tasks suggests that the observed effects reflect an increased cognitive load.
... However, early word recognition processes may occur with little attentional engagement (Dupoux et al., 2003). For instance, dichotic listening studies, where participants are asked to attend to one source of information (e.g. a female voice) while ignoring another source (e.g. a male speaker), have shown that the unattended speech is nonetheless processed to some extent (Aydelott et al., 2015;Dupoux et al., 2003;Rivenez et al., 2006;Rivenez et al., 2008). This means speakers' goals (e.g. ...
Article
Full-text available
In a novel continuous speaking-listening paradigm, we explored how speech planning was affected by concurrent listening. In Experiment 1, Dutch speakers named pictures with high versus low name agreement while ignoring Dutch speech, Chinese speech, or eight-talker babble. Both name agreement and type of auditory input influenced response timing and chunking, suggesting that representational similarity impacts lexical selection and the scope of advance planning in utterance generation. In Experiment 2, Dutch speakers named pictures with high or low name agreement while either ignoring Dutch words, or attending to them for a later memory test. Both name agreement and attention demand influenced response timing and chunking, suggesting that attention demand impacts lexical selection and the planned utterance units in each response. The study indicates that representational similarity and attention demand play important roles in linguistic dual-task interference, and the interference can be managed by adapting when and how to plan speech.
... Ever since then, researchers have sought to explain this phenomenon in terms of psychological models [26][27][28] and neurophysiological data 15,[29][30][31] . Despite these efforts, the extent to which unattended speech is semantically processed by the brain remains unclear 32,33 . However, given the very marked limitations in the ability of subjects to report on the content of unattended speech, it is reasonable to assert that such speech is not processed to the same depth as attended speech. ...
Preprint
Full-text available
Understanding natural speech requires that the human brain convert complex spectrotemporal patterns of acoustic input into meaning in a rapid manner that is reasonably tightly time-locked to the incoming speech signal. However, neural evidence for such a time-locked process has been lacking. Here, we sought such evidence by using a computational model to quantify the meaning carried by each word based on how semantically dissimilar it was to its preceding context and then regressing this quantity against electroencephalographic (EEG) data recorded from subjects as they listened to narrative speech. This produced a prominent negativity at a time-lag of 200– 600 ms on centro-parietal EEG electrodes. Subsequent EEG experiments involving time-reversed speech, cocktail party attention and audiovisual speech-in-noise demonstrated that this response was exquisitely sensitive to whether or not subjects were understanding the speech they heard. These findings demonstrate that, when successfully comprehending natural speech, the human brain encodes meaning as a function of the amount of new information carried by each word in a relatively time-locked fashion.
Article
Many situations require focusing attention on one speaker, while monitoring the environment for potentially important information. Some have proposed that dividing attention among 2 speakers involves behavioral trade-offs, due to limited cognitive resources. However the severity of these trade-offs, particularly under ecologically-valid circumstances, is not well understood. We investigated the capacity to process simultaneous speech using a dual-task paradigm simulating task-demands and stimuli encountered in real-life. Participants listened to conversational narratives (Narrative Stream) and monitored a stream of announcements (Barista Stream), to detect when their order was called. We measured participants’ performance, neural activity, and skin conductance as they engaged in this dual-task. Participants achieved extremely high dual-task accuracy, with no apparent behavioral trade-offs. Moreover, robust neural and physiological responses were observed for target-stimuli in the Barista Stream, alongside significant neural speech-tracking of the Narrative Stream. These results suggest that humans have substantial capacity to process simultaneous speech and do not suffer from insufficient processing resources, at least for this highly ecological task-combination and level of perceptual load. Results also confirmed the ecological validity of the advantage for detecting ones’ own name at the behavioral, neural, and physiological level, highlighting the contribution of personal relevance when processing simultaneous speech.
Article
A wide variety of research and clinical assessments involve presenting speech stimuli in the presence of some kind of noise. Here, I selectively review two theoretical perspectives and discuss ways in which these perspectives may help researchers understand the consequences for listeners of adding noise to a speech signal. I argue that adding noise changes more about the listening task than merely making the signal more difficult to perceive. To fully understand the effects of an added noise on speech perception, we must consider not just how much the noise affects task difficulty, but also how it affects all of the systems involved in understanding speech: increasing message uncertainty, modifying attentional demand, altering affective response, and changing motivation to perform the task.
Article
Users are capable of noticing, listening, and comprehending concurrent information simultaneously, but in conventional speech-based interaction methods, systems communicate information sequentially to the users. This mismatch implies that the sequential approach may be under-utilising human perception capabilities and restricting users to seek information to sub-optimal levels. This paper reports on an experiment that investigates the cognitive workload experienced by the users when listening to a variety of combinations of information types in concurrent formats. Fifteen different combinations of concurrent information streams were investigated, and the subjective listening workload for each of the combination was measured using NASA-TLX. The results showed that the perceived workload index score varies in all concurrent combinations. The workload index score depends on the types and the amount of information presented to users. The perceived workload index score in concurrent listening remained the highest in Monolog with Interview (three concurrent talkers) combination, medium in Monolog with News Headlines (two talkers where one is intermittent) combination, and the lowest in Monolog with Music (one talker and a concurrent music stream) combination. Users descriptive feedback remained aligned with the NASA-TLX-based results. It is expected that the results of this experiment will contribute to helping digital content creators and interaction designers to communicate information more efficiently to users.
Article
In human-computer interaction, particularly in multimedia delivery, information is communicated to users sequentially, whereas users are capable of receiving information from multiple sources concurrently. This mismatch indicates that a sequential mode of communication does not utilise human perception capabilities as efficiently as possible. This article reports an experiment that investigated various speech-based (audio) concurrent designs and evaluated the comprehension depth of information by comparing comprehension performance across several different formats of questions (main/detailed, implied/stated). The results showed that users, besides answering the main questions, were also successful in answering the implied questions, as well as the questions that required detailed information, and that the pattern of comprehension depth remained similar to that seen to a baseline condition, where only one speech source was presented. However, the participants answered more questions correctly that were drawn from the main information, and performance remained low where the questions were drawn from detailed information. The results are encouraging to explore the concurrent methods further for communicating multiple information streams efficiently in human-computer interaction, including multimedia.
Article
The present study investigated how single-talker and babble maskers affect auditory and lexical processing during native (L1) and non-native (L2) speech recognition. Electroencephalogram (EEG) recordings were made while L1 and L2 (Korean) English speakers listened to sentences in the presence of single-talker and babble maskers that were colocated or spatially separated from the target. The predictability of the sentences was manipulated to measure lexical-semantic processing (N400), and selective auditory processing of the target was assessed using neural tracking measures. The results demonstrate that intelligible single-talker maskers cause listeners to attend more to the semantic content of the targets (i.e., greater context-related N400 changes) than when targets are in babble, and that listeners track the acoustics of the target less accurately with single-talker maskers. L1 and L2 listeners both modulated their processing in this way, although L2 listeners had more difficulty with the materials overall (i.e., lower behavioral accuracy, less context-related N400 variation, more listening effort). The results demonstrate that auditory and lexical processing can be simultaneously assessed within a naturalistic speech listening task, and listeners can adjust lexical processing to more strongly track the meaning of a sentence in order to help ignore competing lexical content.
Article
Full-text available
In the present study, we used event-related potentials (ERPs) to examine the effects of prime-target repetition using a dichotic priming paradigm. Participants monitored a stream of target words in the right, attended ear for occasional animal names, and ERPs were recorded to nonanimal words that were either unrelated to or a repetition of prime words presented to the left ear. The prime words were spoken in a different voice and had a lower intensity than did the target words, and the prime word onset occurred 50 ms before target word onset. Repetition-priming effects were observed in the ERPs starting around 150 ms post-target-onset and continued to influence processing for the duration of the target stimuli. These priming effects provide further evidence in favor of parallel processing of overlapping dichotic stimuli, at least up to the level of some form of sublexical phonological representation, a likely locus for the integration of the two sources of information.
Article
Full-text available
We present word frequencies based on subtitles of British television programmes. We show that the SUBTLEX-UK word frequencies explain more of the variance in the lexical decision times of the British Lexicon Project than the word frequencies based on the British National Corpus and the SUBTLEX-US frequencies. In addition to the word form frequencies, we also present measures of contextual diversity part-of-speech specific word frequencies, word frequencies in children programmes, and word bigram frequencies, giving researchers of British English access to the full range of norms recently made available for other languages. Finally, we introduce a new measure of word frequency, the Zipf scale, which we hope will stop the current misunderstandings of the word frequency effect.
Article
Full-text available
Effect sizes are the most important outcome of empirical studies. Most articles on effect sizes highlight their importance to communicate the practical significance of results. For scientists themselves, effect sizes are most useful because they facilitate cumulative science. Effect sizes can be used to determine the sample size for follow-up studies, or examining effects across studies. This article aims to provide a practical primer on how to calculate and report effect sizes for t-tests and ANOVA's such that effect sizes can be used in a-priori power analyses and meta-analyses. Whereas many articles about effect sizes focus on between-subjects designs and address within-subjects designs only briefly, I provide a detailed overview of the similarities and differences between within- and between-subjects designs. I suggest that some research questions in experimental psychology examine inherently intra-individual effects, which makes effect sizes that incorporate the correlation between measures the best summary of the results. Finally, a supplementary spreadsheet is provided to make it as easy as possible for researchers to incorporate effect size calculations into their workflow.
Article
Full-text available
This paper proposed a two-stage model to capture some basic relations between attention, comprehension and memory for sentences. According to the model, the first stage of linguistic processing is carried out in short-term memory (M1) and involves a superficial analysis of semantic and syntactic features of words. The second stage is carried out in long-term memory (M2) and involves application of transformational rules to the analyses of M1 so as to determine the deep or underlying relations among words and phrases. According to the theory, attention is an M2 process: preliminary analyses by M1 are carried out even for unattended inputs, but final analyses by M2 are only carried out for attended inputs. The theory was shown to be consistent with established facts concerning memory, attention and comprehension, and additional support for the theory was obtained in a series of dichotic listening experiments.
Article
The authors used lexical decision in a dichotic listening situation and measured identity priming across channels to explore whether unattended stimuli can be processed lexically. In 6 experiments, temporal synchronization of prime and target words was manipulated, and acoustic saliency of the unattended prime was varied by embedding it in a carrier sentence or in babble speech. When the prime was acoustically salient, a cross-channel priming effect emerged, and participants were aware of the prime. When the prime was less salient, no identity priming was found, and participants failed to notice the prime. Saliency was manipulated in ways that did not degrade the prime. Results are inconsistent with models of late filtering, which predict equal priming irrespective of prime saliency.
Article
The signal-to-noise ratio in typical cocktail-party situations has been calculated and compared with the masked speech reception threshold for sentences. For the free-field condition (no ceiling) the sound level of the voice babble of all competing talkers is only about 2 dB below the limit for satisfactory understanding of the near-by talker one is listening to. In view of smallness of this margin and the unfavourable effect of ambient noise on the vocal output, the intensity of the noise reflected by the ceiling and the walls should not exceed the intensity of the free-field noise. The sound-absorption coefficients of the walls and the ceiling required to meet this condition are given. Since such measures are insufficient for hearing-impaired people (main category: elderly people with presbyacusis), rooms used by them for social gatherings should be highly sound-absorbent.
Article
This paper describes a computerised database of psycholinguistic information. Semantic, syntactic, phonological and orthographic information about some or all of the 98,538 words in the database is accessible, by using a specially-written and very simple programming language. Word-association data are also included in the database. Some examples are given of the use of the database for selection of stimuli to be used in psycholinguistic experimentation or linguistic research. © 1981, The Experimental Psychology Society. All rights reserved.
Article
Broadbent (1956) reports that two lists of digits, each presented to one ear separately so that the items in the two series coincide in time, are grouped together according to the ear-of-arrival, and that these two lists are accordingly recalled separately, one after the other. To ascertain whether such a tendency reflected some built-in mechanism or whether it was due to an optional tactic, adopted through success in making sense of message sequences in other situations, an experiment was designed in which a meaningful message would emerge for the subject if the ear-of-arrival cue was ignored. In this experiment, words broken up into syllables, and phrases broken up into their monosyllabic constituent words were presented to the subject, with the constituents alternating between the two ears. At the same time lists of digits were presented to whichever ear was unoccupied. The results show that recall by meaning rather than by ear-of-arrival, when these are in conflict, can occur and is no less efficient.
Article
Three experiments using dichotic listening measured the priming effect produced on the detection of a semantically defined target word in an attended list of words by a lexically identical word presented to the opposite ear with an attenuation of 12 dB. Experiment 1 embedded the prime in unattended continuous speech and found a 21 ms priming effect, but only when the voices of the two messages had different pitch ranges. Experiment 2 emphasised and generalised the priming effect to a more natural situation using voices differing both in pitch and timbre. This priming effect did not vary with word frequency. In Experiment 3 the prime was part of a list of isolated words played to the unattended ear. It found a 94 ms priming effect when there was no pitch-range difference between the two messages. This priming effect was larger for high-frequency words than for low. These results extend recent findings by Rivenez, Darwin, and Guillaume (2006) and demonstrate the importance of factors that influence perceptual organisation in determining the extent to which unattended messages can be processed.
Article
This paper reviews the evidence relating lateral asymmetry in auditory perception to the asymmetrical functioning of the two hemispheres of the brain. Because each ear has greater neural representation in the opposite cerebral hemisphere, the predominance of the left hemispere for speech is reflected in superior recognition for words arriving at the right ear, while the predominance of the right hemisphere in melodic-pattern perception is reflected in superior identification of melodies arriving at the left ear. Some applications of the dichotic listening technique to questions concerned with the development of cerebral dominance, and with the further specification of function of the left and right hemispheres, are also described.