ArticlePDF Available

Sentence comprehension in competing speech: Dichotic sentence-word priming reveals hemispheric differences in auditory semantic processing

Authors:

Abstract and Figures

This study examined the effects of competing speech on auditory semantic comprehension using a dichotic sentence-word priming paradigm. Lexical decision performance for target words presented in spoken sentences was compared in strongly and weakly biasing semantic contexts. Targets were either congruent or incongruent with the sentential bias. Sentences were presented to one auditory channel (right or left), either in isolation or with competing speech produced by a single talker of the same gender presented simultaneously. The competing speech signal was either presented in the same auditory channel as the sentence context, or in a different auditory channel, and was either meaningful (played forward) or unintelligible (time-reversed). Biasing contexts presented in isolation facilitated responses to congruent targets and inhibited responses to incongruent targets, relative to a neutral baseline. Facilitation priming was reduced or eliminated by competing speech presented in the same auditory channel, supporting previous findings that semantic activation is highly sensitive to the intelligibility of the context signal. Competing speech presented in a different auditory channel affected facilitation priming differentially depending upon ear of presentation, suggesting hemispheric differences in the processing of the attended and competing signals. Results were consistent with previous claims of a right ear advantage for meaningful speech, as well as with visual word recognition findings implicating the left hemisphere in the generation of semantic predictions and the right hemisphere in the integration of newly encountered words into the sentence-level meaning. Unlike facilitation priming, inhibition was relatively robust to the energetic and informational masking effects of competing speech, and was not influenced by the strength of the contextual bias or the meaningfulness of the competing signal, supporting a two-process model of sentence priming in which inhibition reflects later-stage, expectancy-driven strategic processes that may benefit from perceptual reanalysis after initial semantic activation.
Content may be subject to copyright.
This article was downloaded by: [King's College London]
On: 24 October 2011, At: 07:10
Publisher: Psychology Press
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Language and Cognitive Processes
Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/plcp20
Sentence comprehension in
competing speech: Dichotic sentence-
word priming reveals hemispheric
differences in auditory semantic
processing
Jennifer Aydelott
a
, Dinah Baer-Henney
b
, Maciej Trzaskowski
a
c
, Robert Leech
a
d
& Frederic Dick
a
a
Department of Psychological Sciences in the School of
Science, Birkbeck, University of London, UK
b
Department of Linguistics, University of Postdam, Germany
c
MRC Social, Genetic and Developmental Psychiatry Research
Centre, Institute of Psychiatry, Kings College London, UK
d
C3NL, Division of Experimental Medicine, Imperial College
London, UK
Available online: 20 Oct 2011
To cite this article: Jennifer Aydelott, Dinah Baer-Henney, Maciej Trzaskowski, Robert Leech
& Frederic Dick (2011): Sentence comprehension in competing speech: Dichotic sentence-word
priming reveals hemispheric differences in auditory semantic processing, Language and Cognitive
Processes, DOI:10.1080/01690965.2011.589735
To link to this article: http://dx.doi.org/10.1080/01690965.2011.589735
PLEASE SCROLL DOWN FOR ARTICLE
Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-
conditions
This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any
representation that the contents will be complete or accurate or up to date. The
accuracy of any instructions, formulae, and drug doses should be independently
verified with primary sources. The publisher shall not be liable for any loss, actions,
claims, proceedings, demand, or costs or damages whatsoever or howsoever caused
arising directly or indirectly in connection with or arising out of the use of this
material.
Downloaded by [King's College London] at 07:10 24 October 2011
Sentence comprehension in competing speech:
Dichotic sentence-word priming reveals hemispheric
differences in auditory semantic processing
Jennifer Aydelott
1
, Dinah Baer-Henney
2
,
Maciej Trzaskowski
1,3
, Robert Leech
1,4
, and Frederic Dick
1
1
Department of Psychological Sciences in the School of Science,
Birkbeck College, University of London, London, UK
2
Department of Linguistics, University of Potsdam, Potsdam,
Germany
3
MRC Social, Genetic and Developmental Psychiatry Research Centre,
Institute of Psychiatry, King’s College London, London, UK
4
C3NL, Division of Experimental Medicine, Imperial College London,
London, UK
This study examined the effects of competing speech on auditory semantic
comprehension using a dichotic sentence-word priming paradigm. Lexical
decision performance for target words presented in spoken sentences was
compared in strongly and weakly biasing semantic contexts. Targets were either
congruent or incongruent with the sentential bias. Sentences were presented to
one auditory channel (right or left), either in isolation or with competing
speech produced by a single talker of the same gender presented simulta-
neously. The competing speech signal was either presented in the same auditory
channel as the sentence context, or in a different auditory channel, and was
either meaningful (played forward) or unintelligible (time-reversed).
Correspondence should be addressed to Jennifer Aydelott, Department of Psychological
Sciences in the School of Science, Birkbeck College, University of London, Malet Street,
London WC1E 7HX, UK. E-mail: j.aydelott@bbk.ac.uk
This research was supported by grants from the Nuffield Foundation (SGS/01155/G), the
Medical Research Council (G0400341), and the Faculty of Science, Birkbeck, University of
London. The authors would like to thank Germaine Symons, Julia Carnevale, Richard Abbott,
and Katie Alcock for collecting pilot data for this project.
LANGUAGE AND COGNITIVE PROCESSES
0000, 00 (00), 137
#
2011 Psychology Press, an imprint of the Taylor & Francis Group, an Informa business
http://www.psypress.com/lcp http://dx.doi.org/10.1080/01690965.2011.589735
Downloaded by [King's College London] at 07:10 24 October 2011
Biasing contexts presented in isolation facilitated responses to congruent
targets and inhibited responses to incongruent targets, relative to a neutral
baseline. Facilitation priming was reduced or eliminated by competing speech
presented in the same auditory channel, supporting previous findings that
semantic activation is highly sensitive to the intelligibility of the context signal.
Competing speech presented in a different auditory channel affected facilita-
tion priming differentially depending upon ear of presentation, suggesting
hemispheric differences in the processing of the attended and competing
signals. Results were consistent with previous claims of a right ear advantage
for meaningful speech, as well as with visual word recognition findings
implicating the left hemisphere in the generation of semantic predictions and
the right hemisphere in the integration of newly encountered words into the
sentence-level meaning. Unlike facilitation priming, inhibition was relatively
robust to the energetic and informational masking effects of competing speech
and was not influenced by the strength of the contextual bias or the
meaningfulness of the competing signal, supporting a two-process model of
sentence priming in which inhibition reflects later-stage, expectancy-driven
strategic processes that may benefit from perceptual reanalysis after initial
semantic activation.
Keywords: Auditory language comprehension; Semantic priming; Hemispheric
asymmetries; Lexical access; Multitalker environments; Competing speech.
INTRODUCTION
A substantial body of research has documented the adverse effects of a
multitalker environment on speech intelligibility, as well as the factors that
determine the successful identification of words and sentences in the presence
of competing speech (Arons, 1992; Bronkhorst, 2000; Brungart, 2001; Wood
& Cowan, 1995). In addition to the energetic masking effects imposed by
speech as a noise source, a competing speech signal introduces informational
masking due to demands on selective attention and auditory segregation
as well as through phonological and semantic interference. Nevertheless,
listeners are able to make use of the contextual information present in the
attended signal to mitigate the masking effects of competing speech. The
presence of a semantic bias in a spoken sentence context substantially
improves the identification of compatible words in a multitalker babble
(Bilger, Nuetzel, Rabinowitz, & Rzeczkowski, 1984; Hutcherson, Dirks, &
Morgan, 1979; Kalikow, Stevens, & Elliott, 1977), suggesting that higher-
level language processes play a significant role in the identification of speech-
in-speech.
The influence of semantic context on spoken word recognition has been
explored extensively in the psycholinguistics literature. Studies of contextual
priming have demonstrated that words that are congruent with a meaningful
context (a semantically related word or biasing sentence) are recognised
more quickly than incongruent words in tasks such as lexical decision
2 AYDELOTT ET AL.
Downloaded by [King's College London] at 07:10 24 October 2011
(word/nonword judgment) in both the visual and auditory modalities
(Fischler & Bloom, 1979, 1980; McNamara, 2005; Meyer & Schvaneveldt,
1971; Neely, 1991; Schuberth & Eimas, 1977; Stanovich & West, 1983). The
influence of sentence-level meaning on lexical decision performance has been
attributed to a combination of comprehension-specific processes, such as the
activation of compatible features in semantic memory (Kutas & Federmeier,
2000; Schwanenflugel & Lacount, 1988; Schwanenflugel & Shoben, 1985;
Schwanenflugel & White, 1991; Traxler & Foss, 2000), and more general
task-related strategies that are dependent upon the output of semantic
processing but are not necessarily specific to language comprehension
(Ratcliff & McKoon, 1981; Stanovich & West, 1983; West & Stanovich,
1982). In contrast to offline measures of speech intelligibility, such as
accuracy in repetition or written identification tasks, auditory contextual
priming provides a measure of speed of processing of word-level information
that is highly sensitive under ideal listening conditions. However, the effect of
an adverse auditory environment on the online processing of sentence-level
meaning has received little attention in the literature.
The present study brings together parallel findings from speech recogni-
tion and psycholinguistics by exploring the effects of a competing speech
signal on auditory contextual priming. The aim of the study is to examine the
different types of interference imposed by a speech-in-speech environment,
and the ways in which these factors affect the activation of representations in
semantic memory and the use of semantic information in the service of task
performance.
Rationale and predictions
As noted above, a competing speech signal has the potential to interfere with
spoken language comprehension in several ways. First, competing speech
imposes energetic masking on the attended speech signal due to the spectral
and temporal overlap between the two signals, thereby reducing the
intelligibility of the attended signal (Brungart, 2001). This effect is
substantially reduced when the attended and competing signals are presented
to separate auditory channels (Cherry, 1953; Freyman, Balakrishnan, &
Helfer, 2001). Further, competing speech produces informational masking to
varying degrees depending on the acoustic similarity of the attended and
competing signals (e.g., competing speech produced by a talker of the same
gender produces greater interference than competing speech produced by a
talker of a different gender; Brungart, 2001). The meaningful content of the
competing signal also contributes to informational masking. For example,
competing speech in a listeners native language is a more effective masker
than a foreign language (Garcia Lecumberri & Cooke, 2006; Van Engen &
Bradlow, 2007), and competing multitalker babble containing high frequency
AUDITORY SENTENCE PRIMING IN COMPETING SPEECH 3
Downloaded by [King's College London] at 07:10 24 October 2011
words has a greater effect on word recognition than babble containing low
frequency words (Boulenger, Hoen, Ferragne, Pellegrino, & Meunier, 2010).
In addition, competing speech may have differential effects depending on the
ear of presentation of the attended and competing signals. Dichotic listening
studies have revealed a right-ear advantage (REA) for the identification of
speech stimuli, which is generally attributed to the preponderance of
contralateral neural connections from the right ear to the left auditory
cortex, in combination with left hemisphere specialisation for speech
processing (Kimura, 1961; Studdert-Kennedy & Shankweiler, 1970; Wada
& Rasmussen, 1960; cf. Scott, Blank, Rosen, & Wise, 2000). Thus, a
competing speech signal presented to the right ear may produce greater
interference than the same signal presented to the left ear.
These potential sources of interference have particular implications for
spoken word recognition in a semantic context. Priming studies have
demonstrated that a biasing sentence context affects lexical decision
response times (RTs) to target words, such that words that are congruent
with the sentence meaning are facilitated and words that are incongruent
with the sentence meaning are inhibited, relative to a neutral baseline
(Aydelott & Bates, 2004; Stanovich & West, 1983). The facilitation of
congruent words in a biasing sentence has been described as reflecting the
activation of a set of features in the semantic system (Schwanenflugel &
LaCount, 1988; cf. Kutas & Federmeier, 2000), and is highly sensitive to the
intelligibility of the speech signal (Aydelott & Bates, 2004). Thus, the
activation of semantic information is likely to be particularly vulnerable to
the energetic masking effect of competing speech on an auditory sentence
context, resulting in reduced facilitation effects. On the other hand, the
inhibition of lexical decision responses to words that are incongruent with
the sentence meaning has been attributed to task-specific strategies, as
targets that are inconsistent with the contextual bias are perceived as
anomalous and tend to trigger a ‘‘nonword’’ response, resulting in slower
RTs when the target is a real word (Neely & Keefe, 1989; Ratcliff &
McKoon, 1981; Stanovich & West, 1983; West & Stanovich, 1982; but cf.
Plaut & Booth, 2000, 2006). Inhibition effects are associated with increased
demands on attentional resources and are reduced or eliminated under
challenging listening conditions in which intelligibility is relatively unaf-
fected, such as temporal compression (Aydelott & Bates, 2004). Thus, the
increased demands on auditory segregation imposed by a competing speech
signal with similar acoustic properties to the attended signal are likely to
interfere with the inhibition of incongruent targets, particularly if the
competing signal is meaningful. In addition, the extent of the interference
produced by competing speech may also depend upon the strength of
4 AYDELOTT ET AL.
Downloaded by [King's College London] at 07:10 24 October 2011
the contextual bias. The support provided by a strongly constraining
semantic context in which the target word is highly predictable substantially
improves target intelligibility when sentences are presented in a multitalker
babble (Bilger et al., 1984; Hutcherson et al., 1979; Kalikow et al., 1977).
The masking effect of competing speech may therefore be less pronounced
for sentences with a strong bias in favour of a particular word.
Design
The present study tested these predictions using a modified version of the
auditory sentence priming paradigm. Participants made lexical decision
responses to spoken word targets (e.g., desk) presented in three auditory
sentence context conditions: strong bias, in which the congruent target was
the expected completion of the sentence (e.g., In the office, the computer is on
my ...); weak bias, in which the congruent target was one of many plausible
completions (e.g., They took all of the furniture in the office except for the ...);
and neutral, in which there was no semantic bias (The next item is ...).
Targets were either congruent or incongruent with the biasing contexts so
that facilitation and inhibition effects could be compared. Both the biasing
and neutral contexts were presented dichotically to a single auditory channel
and were played either in isolation or with competing speech produced by a
single talker of the same gender as the talker who produced the context
stimuli.
In all competing speech conditions, the critical experimental questions
were: (1) whether the competing speech signal significantly reduced the
facilitation of congruent targets and/or the inhibition of incongruent targets,
relative to conditions in which the attended sentence contexts were presented
in isolation; and (2) whether this reduction was modulated by the contextual
strength of the attended sentence. Thus, the presence or absence of a
competing speech signal and the strength of the contextual bias were treated
as within-subjects variables in a repeated measures design.
The spatial location and information content of the competing signal were
manipulated independently as between-subjects variables. The following
factors were examined:
Energetic masking/binaural release from masking
The competing speech signal was either mixed with the sentence context
and presented in the same auditory channel to maximise its energetic
masking effect, or presented in a different auditory channel from the
sentence context to maximise the potential benefit from binaural spatial
AUDITORY SENTENCE PRIMING IN COMPETING SPEECH 5
Downloaded by [King's College London] at 07:10 24 October 2011
unmasking. Energetic masking was expected to reduce facilitation effects for
congruent targets, particularly in weakly biasing sentence contexts.
Meaningfulness of competing speech
The competing speech masker was either played forward or time-reversed.
Time-reversed speech has similar spectral and temporal properties to
forward speech but lacks semantic content. Forward speech was expected
to increase the informational masking effect of the competing signal, placing
greater demands on attentional resources relative to time-reversed speech,
and was therefore expected to have a greater effect on the inhibition of
incongruent targets. Forward competing speech was also expected to reduce
facilitation due to the activation of incompatible representations in semantic
memory.
Ear of presentation
The competing signal was either presented to the right or to the left
auditory channel. Meaningful competing speech presented to the right
auditory channel was expected to be more disruptive than meaningful speech
presented to the left channel, due to the REA for intelligible speech observed
in previous studies.
In Experiment 1, all competing speech signals were presented at a
moderate signal-to-noise ratio (SNR) of 0 dB. A more demanding SNR of
12 dB is used in Experiment 2.
EXPERIMENT 1
Method
Stimuli
All targets, sentence contexts, and competing speech segments were
recorded onto digital audio tape in an Industrial Acoustics Corporation
403-A audiometric chamber with a Tascam DA-P1 tape recorder and a
Sennheiser ME65/K6 supercardioid microphone and pre-amp at gain levels
between 6 and 12 dB. The recorded stimuli were transferred via digital-
to-digital sampling onto a Macintosh G4 computer with a Digidesign audio
card using ProTools LE software at a sampling rate of 44.1 kHz with a 16-bit
quantisation. The waveform of each target item, sentence context, and
competing speech segment was edited and saved in its own mono audio file in
WAV format for subsequent manipulation using Praat software (Boersma,
2001).
6 AYDELOTT ET AL.
Downloaded by [King's College London] at 07:10 24 October 2011
Target items
Word targets were 60 one-syllable English words (see Appendix) contain-
ing three to five phonemes (mean 3.30, SD0.65) with a mean duration
of 643 ms (SD87), a mean Kucera-Francis print frequency of 139
(SD104; Kucera & Francis, 1967), a mean London-Lund spoken
frequency of 17 (SD25; Brown, 1984), and a mean concreteness rating
of 536 (SD87; all values obtained from the MRC Psycholinguistic
Database; Coltheart, 1981). To avoid possible morphological and morpho-
phonological constraints on determiners (a/an, the), mass nouns (e.g., blood,
dust) were excluded, and all targets were consonant-initial. Nonword
distractor targets consisted of 60 phonologically permissible one-syllable
nonsense items that did not differ significantly from the word targets in
terms of number of phonemes (mean 3.38, SD0.58) or duration in ms
(mean 666, SD112). All target items were produced by a male native
speaker of Southern British English, and the resulting waveforms were scaled
to a nominal average intensity level of 72 dB in Praat.
Sentence contexts
Two sentence contexts were created for each target word: one with a
strong semantic bias in favour of the target, and one with a weak semantic
bias in favour of the target (see Appendix). The predictability of the intended
target word in each of these sentence contexts was then evaluated by 22
native speakers of British English using the cloze procedure. Target words
had a mean cloze probability of 94% (SD7%) in strong bias contexts and
28% (SD19%) in weak bias contexts. Strong and weak bias sentences did
not differ significantly in number of syllables (strong: mean 10.03,
SD2.70; weak: mean 9.98, SD2.80), duration in ms (strong:
mean 2036, SD477; weak: mean 2065, SD470), or number of
content words (strong: mean 3.65, SD1.31; weak: mean 3.40,
SD1.17); however, strong bias sentences contained significantly more
words that were semantically related to the target item than weak bias
sentences, strong: mean 1.15, SD0.61; weak: mean 0.87, SD0.83;
paired t(59) 3.43, pB.01. A sentence context without a semantic bias in
favour of a particular target (The next item is ...) served as the neutral
baseline.
Distractor sentence contexts were also generated for the nonword targets.
These were strongly or weakly biased in favour of words that were not
included in the target set (cloze probability of expected word, strong:
mean 71%, SD24%; weak: mean 41%, SD31%), although they
were always presented with nonword targets. Distractor sentence contexts
did not differ significantly from test sentence contexts in number of syllables
AUDITORY SENTENCE PRIMING IN COMPETING SPEECH 7
Downloaded by [King's College London] at 07:10 24 October 2011
(strong: mean 10.13, SD2.83; weak: mean 9.73, SD2.67), duration
in ms (strong: mean 2188, SD 513; weak: mean 1995, SD485),
or number of content words (strong: mean 3.80, SD1.39; weak:
mean 3.47, SD1.23).
All sentence contexts were produced by a female native speaker of
Southern British English and the resulting waveforms were scaled to a
nominal average intensity level of 72 dB in Praat.
Competing speech
A passage from the economics textbook Profit Patterns (Slywotzky, 1999)
was read aloud by a different female native speaker of Southern British
English. Segments of the same duration as each of the sentence contexts were
excised at random from this recorded passage, with onset and offset cuts
made at zero crossings of the waveform. These excised portions were scaled
to a nominal average intensity level of 72 dB in Praat.
Dichotic sentence stimuli
For each of the sentence contexts, a stereo sound file was generated with
the sentence waveform inserted into the left channel and silence in the right
channel. These stereo files served as the isolation condition.
A time-reversed version of each of the excised competing speech segments
was created in Praat. Separate dichotic sentence stimuli were generated for
forward and time-reversed competing speech segments. Each competing
speech segment was combined with its duration-matched sentence context in
the left channel of the stereo sound file (same channel competing speech
condition), or inserted into the right channel (different channel competing
speech condition).
All stimuli (dichotic sentence contexts and targets) were converted in
SoundEdit 16 to System 7 format for presentation via SuperLab software.
Participants
Eighty right-handed native speakers of British English between the ages of
18 and 40 were paid for their participation. Individuals with a history of
hearing impairment or neurological illness were excluded from participating.
Procedure
Sentence contexts were paired with word targets in five semantic bias
conditions: congruent strong (target was an expected completion of a strong
bias context), congruent weak (target was an expected completion of a weak
bias context), neutral (target was presented in a neutral context), incongruent
8 AYDELOTT ET AL.
Downloaded by [King's College London] at 07:10 24 October 2011
strong (target was an unexpected completion of a strong bias context), and
incongruent weak (target was an unexpected completion of a weak bias
context). For the incongruent conditions, the same sentence contexts were
used as in the congruent conditions, paired with different (unexpected)
targets. Thus, the sentence contexts and targets served as their own controls,
as the same test items were used in both the congruent and incongruent
conditions. In each of the semantic bias conditions, the sentence context was
either presented in isolation, or with competing speech. For each participant,
each context and target appeared only once during the experiment, but the
condition in which each context and target appeared was counterbalanced
across participants. Thus, all items were presented in all conditions, and all
participants received all items, obviating the need to conduct separate
analyses by item (McNamara, 2005). Semantic bias (congruent strong,
congruent weak, neutral, incongruent strong, or incongruent weak) and
competing signal (isolation versus competing speech) served as within-
subjects variables. The type of competing signal (forward versus time-
reversed speech), the auditory channel in which the competing signal was
presented with respect to the context sentence (same channel versus different
channel), and the ear of presentation of the context sentence (left versus
right) served as between-subjects variables. Nonwords were presented in the
same context conditions as word targets. Target words and nonwords were
always presented binaurally, with no competing speech.
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 an Industrial Acoustics
Corporation 403-A audiometric chamber. Ear of presentation of the sentence
context was determined for each participant by the placement of the
reversible headphones, such that the left channel (i.e., the channel containing
the sentence context) was positioned over either the left or the right ear. The
RTs and accuracy were recorded in SuperLab using a Cedrus RB-730
response box. Participants were instructed to listen to the sentence context
and target item and to indicate whether the target was a real word in English
or not by pressing the green button (yes) or the red button (no) on the
response box. Participants were informed that the sentence context would
sometimes be presented with competing speech and were directed to ignore
the competing speech and pay attention to the sentence context. 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 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 as both their response times and accuracy would be
measured.
AUDITORY SENTENCE PRIMING IN COMPETING SPEECH 9
Downloaded by [King's College London] at 07:10 24 October 2011
Results
One participant responded correctly to fewer than 60% of word targets in
one of the experimental conditions and was excluded from analysis. For the
remaining participants, harmonic mean RTs were calculated for correct
responses to word targets in each experimental condition. The harmonic
mean is based on the inverse transformation, which is relatively robust to
outliers in lexical decision RT data (Ratcliff, 1993; cf. Tyler, Randall, &
Marslen-Wilson, 2002). This value was divided by the proportion of correct
responses in each experimental condition to compute the inverse efficiency
(IE) score, a combined measure of RT and accuracy that adjusts for speed-
accuracy trade-offs (Townsend & Ashby, 1983).
The effect of competing speech on performance in the neutral baseline
condition was tested in a four-way ANOVA performed on the IE scores, with
competing signal (isolation versus competing speech) as a within-subjects
variable and type of competing signal (forward versus time-reversed speech),
channel of presentation of competing signal (same or different channel from
sentence contexts), and ear of presentation of sentence contexts (left versus
right) as between-subjects variables. The results revealed no significant main
effect of competing signal on neutral IE scores, F(1, 71) 1.54, p.22, and
no significant interactions.
The IE scores were used to calculate the proportion priming for targets
in each of the semantic bias conditions relative to the neutral baseline (the
difference between the IE scores in the bias and neutral conditions,
divided by the neutral IE score; cf. Burke & Yee, 1984; Tyler et al., 2002).
Mean proportion priming scores across experimental conditions are shown
for congruent targets in Table 1 and for incongruent targets in Table 2.
Separate analyses examined the effects of competing speech in the same
auditory channel as the attended sentence context (reported as Experiment
1A, N39), and competing speech in a different auditory channel from
the attended sentence context (reported as Experiment 1B, N40) on
facilitation and inhibition priming. As none of the critical predictions
relied on a direct comparison of the relative magnitudes of facilitation and
inhibition priming and to simplify the interpretation of the observed
effects, responses to congruent and incongruent targets were analysed
separately.
Experiment 1A: Competing speech in the same channel
Four-way analyses of variance (ANOVAs) with bias strength (strong
versus weak) and competing signal (isolation versus competing speech) as
within-subjects variables and type of competing signal (forward versus
10 AYDELOTT ET AL.
Downloaded by [King's College London] at 07:10 24 October 2011
TABLE 1
Mean (and standard error) proportion priming for congruent targets in biasing
contexts relative to neutral baseline in Experiment 1 (0 dB SNR), based on inverse
efficiency scores
Type of
competing signal
Channel of
presentation
(competing
signal)
Ear of
presentation
(context)
Presence of
competing signal
Bias
strength
Proportion
priming
Forward Same Left Isolation Strong 0.18 (0.033)
Weak 0.13 (0.034)
Competing speech Strong 0.06 (0.031)
Weak 0.02 (0.028)
Right Isolation Strong 0.14 (0.034)
Weak 0.09 (0.036)
Competing speech Strong 0.08 (0.032)
Weak 0.01 (0.030)
Different Left Isolation Strong 0.18 (0.033)
Weak 0.09 (0.034)
Competing speech Strong 0.05 (0.031)
Weak 0.01 (0.028)
Right Isolation Strong 0.12 (0.033)
Weak 0.01 (0.034)
Competing speech Strong 0.12 (0.031)
Weak 0.05 (0.028)
Time-reversed Same Left Isolation Strong 0.16 (0.033)
Weak 0.10 (0.034)
Competing speech Strong 0.05 (0.031)
Weak 0.02 (0.028)
Right Isolation Strong 0.14 (0.033)
Weak 0.05 (0.034)
Competing speech Strong 0.07 (0.031)
Weak 0.03 (0.028)
Different Left Isolation Strong 0.16 (0.033)
Weak 0.07 (0.034)
Competing speech Strong 0.12 (0.031)
Weak 0.10 (0.028)
Right Isolation Strong 0.08 (0.033)
Weak 0.05 (0.034)
Competing speech Strong 0.10 (0.031)
Weak 0.03 (0.028)
Note: Values less than zero reflect facilitation.
AUDITORY SENTENCE PRIMING IN COMPETING SPEECH 11
Downloaded by [King's College London] at 07:10 24 October 2011
time-reversed speech) and ear of presentation (left versus right) as
between-subjects variables were conducted separately for congruent and
incongruent targets.
TABLE 2
Mean (and standard error) proportion priming for incongruent targets in biasing
contexts relative to neutral baseline in Experiment 1 (0 dB SNR), based on inverse
efciency scores
Type of
competing signal
Channel of
presentation
(competing
signal)
Ear of
presentation
(context)
Presence of
competing signal
Bias
strength
Proportion
priming
Forward Same Left Isolation Strong 0.22 (0.053)
Weak 0.18 (0.074)
Competing speech Strong 0.10 (0.054)
Weak 0.10 (0.056)
Right Isolation Strong 0.24 (0.056)
Weak 0.34 (0.078)
Competing speech Strong 0.08 (0.057)
Weak 0.15 (0.059)
Different Left Isolation Strong 0.14 (0.053)
Weak 0.16 (0.074)
Competing speech Strong 0.18 (0.054)
Weak 0.13 (0.056)
Right Isolation Strong 0.25 (0.053)
Weak 0.19 (0.074)
Competing speech Strong 0.25 (0.054)
Weak 0.16 (0.056)
Time-reversed Same Left Isolation Strong 0.13 (0.053)
Weak 0.10 (0.074)
Competing speech Strong 0.18 (0.054)
Weak 0.06 (0.056)
Right Isolation Strong 0.16 (0.053)
Weak 0.24 (0.074)
Competing speech Strong 0.10 (0.054)
Weak 0.12 (0.056)
Different Left Isolation Strong 0.08 (0.053)
Weak 0.16 (0.074)
Competing speech Strong 0.09 (0.054)
Weak 0.10 (0.056)
Right Isolation Strong 0.17 (0.053)
Weak 0.13 (0.074)
Competing speech Strong 0.17 (0.054)
Weak 0.24 (0.056)
Note: Values greater than zero reflect inhibition.
12 AYDELOTT ET AL.
Downloaded by [King's College London] at 07:10 24 October 2011
Congruent targets. Results revealed significant main effects of both bias
strength, F(1, 35) 29.85, pB.0001, and competing signal, F(1, 35) 15.20,
pB.0001. Thus, strong contexts produced greater facilitation than weak
contexts, and facilitation of congruent targets was reduced when sentence
contexts were presented in competing speech relative to isolation. These data
are plotted in Figure 1. There were no other significant main effects and no
-0.3
-0.25
-0.2
-0.15
-0.1
-0.05
0
weak bias
strong bias
weak bias
strong bias
Proportion Priming
isolation competing speech
0
0.05
0.1
0.15
0.2
0.25
0.3
Proportion Priming
* *
* *
isolation competing speech
Figure 1. Effect of competing speech presented in the same auditory channel as the sentence
context on the facilitation of congruent targets (top panel) and inhibition of incongruent targets
(bottom panel) in Experiment 1 (0 dB SNR). Values reflect proportion priming relative to the
neutral baseline based on inverse efficiency scores. In this and subsequent figures, * significant
comparison, pB.05; trend, p B.10; ns comparison not significant.
AUDITORY SENTENCE PRIMING IN COMPETING SPEECH 13
Downloaded by [King's College London] at 07:10 24 October 2011
interactions. One-sample t-tests were conducted to establish whether signifi-
cant facilitation relative to the neutral baseline was obtained in all
experimental conditions. Significant facilitation emerged for sentences pre-
sented in isolation for both strong, t(38) 9.11, pB.0001, and weak
contexts, t(38) 5.68, pB.0001, as well as for strong contexts presented
in competing speech, t(38) 3.74, p B.001; however, no significant
facilitation emerged for weak contexts presented in competing speech,
t(38) 0.91, p.37.
Incongruent targets. A significant main effect of competing signal, F(1,
35) 6.15, pB.05, emerged, such that inhibition of incongruent targets was
reduced when sentence contexts were presented in competing speech relative
to the same contexts presented in isolation. One-sample t-tests revealed that
significant inhibition emerged for sentence contexts presented in isolation,
strong: t(38) 6.04, pB.0001; weak: t(38) 4.77, pB.0001, and in
competing speech, strong: t(38) 4.28, pB.0001; weak: t(38) 4.02,
pB.0001. However, the main effect of bias strength (strong versus weak)
did not approach significance, F(1, 35) 0.10, p.76. Thus, unlike
facilitation effects, inhibition of incongruent targets was not significantly
influenced by the strength of the semantic bias imposed by the context.
These data are plotted in Figure 1. A significant interaction of bias
strengthear of presentation, F(1, 35) 5.35, pB.05, emerged. However,
when this interaction was examined in separate three-way ANOVAs (bias
strengthcompeting signal type of competing signal) conducted for each
ear of presentation, the main effect of bias strength did not reach
significance in either condition, left ear: F(1, 18) 2.62, p.12; right
ear: F(1, 17) 2.74, p.12. There were no other significant main effects or
interactions.
Experiment 1B: Competing speech in a different channel
Four-way ANOVAs with bias strength and competing signal as within-
subjects variables and type of competing signal and ear of presentation as
between-subjects variables were conducted separately for congruent and
incongruent targets.
Congruent targets. Results revealed a significant interaction of ear of
presentationcompeting signal, F(1, 36) 5.25, pB.05, and a marginally
significant interaction of ear of presentationcompeting signaltype of
competing signal, F(1, 36) 3.86, p.057. These findings suggest that
hemispheric processing differences may have contributed to the observed
effects. To clarify the nature of these interactions, separate three-way
ANOVAs with bias strength and competing signal as within-subjects
variables and ear of presentation as a between-subjects variable were
14 AYDELOTT ET AL.
Downloaded by [King's College London] at 07:10 24 October 2011
conducted for each type of competing signal (forward and time-reversed
competing speech, N 20 in each group). For the sake of brevity, ear of
presentation effects are described as reflecting processing in the contralateral
hemisphere, which receives the preponderance of the auditory input,
although perceptual processing in the auditory modality also reflects
ipsilateral input from a minority of ascending pathways.
Forward competing speech. The main effect of bias strength was
significant, F(1, 18) 24.15, pB.0001, such that greater facilitation was
obtained in strong contexts than in weak contexts. A marginally significant
main effect of competing signal also emerged, F(1, 18) 3.51, p.077,
indicating that facilitation was reduced by competing speech. In addition,
there was a significant interaction of ear of presentationcompeting signal,
F(1, 18) 6.56, p B.05, suggesting that the effect of forward competing
speech on facilitation differed depending upon whether the attended sentence
context was presented to the left or the right ear. Paired t-tests revealed that,
when the sentence context was presented to the left ear (right hemisphere)
forward competing speech significantly reduced facilitation for both strong,
t(9) 3.71, pB.05, and weak contexts, t(9) 2.74, pB.05, relative to the
isolation (no competing speech) condition. However, when the sentence
context was presented to the right ear (left hemisphere), forward competing
speech had no significant effect on facilitation relative to the isolation
condition, strong contexts: t(9) 0.17, p.87; weak contexts: t(9) 0.64,
p.54. These data are plotted in Figure 2.
Time-reversed competing speech. A significant competing signal bias
strengthear of presentation interaction emerged, F(1, 18) 5.30, pB.05.
Separate two-way ANOVAs were conducted for each ear of presentation.
Results revealed a significant main effect of bias strength for sentences
presented to the left ear (right hemisphere), F(1, 9) 12.19, pB.01, and a
marginally significant main effect of bias strength for sentences presented to
the right ear (left hemisphere), F(1, 9) 4.95, p .053. The main effect of
competing signal was not significant, nor was there a significant competing
signalbias strength interaction for either ear of presentation. These data
are plotted in Figure 2.
Incongruent targets. There were no significant main effects or interac-
tions. One-sample t-tests revealed that significant inhibition emerged for
sentence contexts presented in isolation, strong: t(39) 7.29, p B.0001;
weak: t(39) 5.55, pB.0001, and in competing speech, strong: t(39)
6.27, p B.0001; weak: t(39) 5.29, pB.0001. These data are plotted in
Figure 3.
AUDITORY SENTENCE PRIMING IN COMPETING SPEECH 15
Downloaded by [King's College London] at 07:10 24 October 2011
Discussion
When presented in isolation, biasing sentence contexts both facilitated
responses to congruent targets and inhibited responses to incongruent
targets, relative to the neutral baseline. Significant facilitation was obtained
for congruent targets in both strong and weak contexts, with greater
facilitation priming emerging for targets in strong contexts than in weak
contexts. This result is consistent with previous studies showing enhanced
-0.3
-0.25
-0.2
-0.15
-0.1
-0.05
0
LH Sentence
RH Sentence
LH Sentence
RH Sentence
Proportion Priming
isolation competing speech
-0.3
-0.25
-0.2
-0.15
-0.1
-0.05
0
Proportion Priming
ns ns
* *
ns ns ns ns
weak biasstrong bias weak biasstrong bias
weak biasstrong bias weak biasstrong bias
isolation competing speech
Figure 2. Effect of forward (top panel) and time-reversed (bottom panel) competing speech in
a different auditory channel from the sentence context, presented to the left hemisphere/right ear
(left plots) or the right hemisphere/left ear (right plots) on facilitation of congruent targets in
Experiment 1 (0 dB SNR).
16 AYDELOTT ET AL.
Downloaded by [King's College London] at 07:10 24 October 2011
facilitation of congruent targets in the presence of a strong semantic bias,
which has been attributed to the degree of constraint imposed by the context:
whereas both strongly and weakly biasing contexts activate compatible
features in semantic memory, strongly biasing contexts also limit the set of
plausible completions, such that compatible words are more easily integrated
(Federmeier, 2007; Wlotko & Federmeier, 2007). Thus, as in the present
study, both high- and low-predictability congruent targets are facilitated,
with additional priming emerging for high-predictability targets. In contrast,
inhibition of incongruent targets was obtained in both strong and weak
contexts, with no difference in the magnitude of inhibition priming due to the
strength of the contextual bias. This suggests that the task-specific strategies
that contribute to the inhibition effect rely upon broad expectancies
generated in response to the activation of semantic features and are not
sensitive to contextual constraint.
Competing speech presented in the same auditory channel as the sentence
context reduced the facilitation of congruent targets in strong contexts and
eliminated the facilitation of congruent targets in weak contexts, supporting
the prediction that sentences with a strong semantic bias would be relatively
robust to reductions in intelligibility introduced by competing speech at this
SNR. The inhibition of incongruent targets was significantly reduced but not
eliminated in both strong and weak contexts, and there was no influence of
bias strength on the effect of competing speech on inhibition priming. Thus,
as in previous studies, inhibition priming was relatively preserved in
conditions of low intelligibility. The meaningfulness of the competing signal
did not affect the pattern of results for facilitation or inhibition priming.
Competing speech presented in a different auditory channel from the
sentence context had differing effects on facilitation depending upon the ear
0
0.05
0.1
0.15
0.2
0.25
0.3
Proportion Priming
isolation competing speech
ns ns
weak bias
strong bias
Figure 3. Effect of competing speech in a different auditory channel from the sentence context
on inhibition of incongruent targets in Experiment 1 (0 dB SNR).
AUDITORY SENTENCE PRIMING IN COMPETING SPEECH 17
Downloaded by [King's College London] at 07:10 24 October 2011
of presentation of the context and the intelligibility of the competing signal.
Forward speech significantly reduced facilitation when the attended sentence
context was presented to the LE/RH, but had no effect when the attended
context was presented to the RE/LH. As predicted, this result is consistent
with the right ear advantage for intelligible speech observed in previous
studies (Kimura, 1961; Studdert-Kennedy & Shankweilier, 1970; Wada &
Rasmussen, 1960): a meaningful speech signal presented to the RE/LH
interferes with the processing of a sentence context presented to the LE/RH
due to the preferential processing of intelligible speech in the LH (cf. Scott
et al., 2000), whereas no such interference is obtained when the context is
presented to the RE/LH. This effect emerged even though participants were
instructed to focus on the sentence context, whereas previous studies have
demonstrated that selective attention can reduce or reverse the REA
(Hugdahl, 2005). Time-reversed speech did not reduce facilitation for either
ear of presentation, indicating that it is the meaningful content of the speech
signal that produces interference, rather than the increased demands imposed
on spatial selective attention.
In contrast to facilitation priming, the inhibition of incongruent targets
was unaffected by competing speech in a different auditory channel,
irrespective of competing signal intelligibility or ear of presentation. This
finding is inconsistent with the prediction that the task-dependent strategies
underlying inhibition priming incur a processing cost and should therefore be
vulnerable under conditions of increased attentional demand. However, it is
possible that the moderate SNR of 0 dB in this experiment was not
sufficiently demanding to interfere with strategic processing. This may also
account for the failure of selective attention to reverse the REA observed for
facilitation priming in competing speech, as the competing signal may not
have induced focused attention on the sentence context at this intensity level.
Alternatively, the presentation of the attended and competing signals at the
same intensity level may have interfered with the perceptual isolation of
the two signals, as interaural intensity differences have been shown to
contribute to auditory segregation in multitalker environments, even when
the target signal is presented at a lower intensity level than the masker
(Brungart, 2001). These possibilities are explored in Experiment 2, which
examines the effect of competing speech presented at a more demanding
SNR of 12 dB on auditory sentence priming using the same materials as in
Experiment 1.
EXPERIMENT 2
Experiment 2 tested the effect of a more challenging SNR on the processing
of words in a semantic context in the presence of competing speech.
18 AYDELOTT ET AL.
Downloaded by [King's College London] at 07:10 24 October 2011
Competing speech presented at an SNR of 12 dB was expected to have an
increased masking effect when presented in the same auditory channel as the
attended signal relative to an SNR of 0 dB, resulting in larger reductions in
both facilitation and inhibition priming. Competing speech presented in a
different auditory channel was expected to place greater demands on
selective attention at the 12 dB than the 0 dB SNR, resulting in reduced
inhibition priming. However, the predictions for facilitation priming in this
condition were less straightforward. Whereas the increased demands of the
more challenging SNR might interfere with contextual facilitation, this SNR
condition was also expected to increase the focus of attention on the attended
signal and to provide an additional interaural difference cue to aid in the
separation of the attended and competing signals. These factors may serve to
override the REA for intelligible speech, eliminating the interference effect
for meaningful competing speech presented to the right auditory channel.
Method
Stimuli
All stimuli were identical to those used in Experiment 1, with the
exception that all sentence contexts (biasing and neutral) were scaled to a
nominal average intensity of 60 dB in Praat (as opposed to 72 dB as in
Experiment 1).
Participants
Eighty native speakers of British English between the ages of 18 and 40
were paid for their participation. Individuals with a history of hearing
impairment or neurological illness were excluded from participating.
Procedure
The procedure was identical to that used in Experiment 1.
Results
Five participants responded correctly to fewer than 60% of word targets in
one or more of the experimental conditions and were excluded from analysis.
Data from two other participants could not be used due to equipment failure.
As in Experiment 1, for the remaining participants, the harmonic mean RT
was calculated for correct responses to word targets in each experimental
condition, and this value was divided by the proportion of correct responses
in each experimental condition to compute the IE score. A four-way ANOVA
performed on IE scores in the neutral baseline condition, with competing
signal as a within-subjects variable and type of competing signal, channel of
AUDITORY SENTENCE PRIMING IN COMPETING SPEECH 19
Downloaded by [King's College London] at 07:10 24 October 2011
presentation, and ear of presentation as between-subjects variables, revealed
no significant main effect of competing signal on neutral IE scores, F(1,
65) 0.94, p .34, and no significant interactions. As in Experiment 1, IE
scores were used to calculate the proportion priming for targets in each of the
semantic bias conditions relative to the neutral baseline. Mean proportion
priming scores across experimental conditions are shown for congruent
targets in Table 3, and for incongruent targets in Table 4. Separate analyses
examined the effects of competing speech in the same auditory channel as
the attended sentence context (reported as Experiment 2A, N 37) and
competing speech in a different auditory channel from the attended sentence
context (reported as Experiment 2B, N36) on facilitation and inhibition
priming.
Experiment 2A: Competing speech in the same channel
Four-way ANOVAs with bias strength and competing signal as within-
subjects variables and type of competing signal and ear of presentation as
between-subjects variables were conducted separately for congruent and
incongruent targets.
Congruent targets. Results revealed significant main effects of both bias
strength, F(1, 33) 4.52, p B.05, and competing signal, F(1, 33) 29.29,
pB.0001. Thus, greater facilitation was obtained for strong contexts than
weak contexts, and competing speech reduced facilitation relative to the
isolation condition. In addition, a significant bias strengthcompeting
signal interaction was obtained, F(1, 33) 15.52, pB.001, such that
significantly greater priming emerged for strong contexts than for weak
contexts in the isolation condition, paired t(36) 4.79, p B.0001, whereas
no significant difference in priming between strong and weak contexts
emerged in competing speech, paired t(36) 0.66, p.50. These data are
plotted in Figure 4. One-sample t-tests revealed that significant priming
was obtained for both strong, t(36) 7.21, p B.0001, and weak,
t(36) 4.33, pB.0001 contexts presented in isolation, and that priming
was eliminated for contexts presented in competing speech, strong:
t(36) 1.05, p .30; weak: t(36) 0.41, p.68. No other significant main
effects or interactions were observed.
Incongruent targets. A significant main effect of competing signal was
obtained, F(1, 33) 32.98, pB.0001, such that smaller inhibition priming
emerged for contexts presented in competing speech than in isolation. These
data are plotted in Figure 4. One-sample t-tests revealed highly significant
inhibitory priming for contexts presented in isolation, strong: t(36) 7.36,
pB.0001; weak: t(36) 6.44, pB.0001, and for strong contexts presented in
competing speech, t(36) 2.09, pB.05. Although significant inhibition was
20 AYDELOTT ET AL.
Downloaded by [King's College London] at 07:10 24 October 2011
not obtained in weak contexts, t(36) 1.13, p.27, the interaction of
competing signalbias strength did not approach significance, F(1,
33) 0.26, p.61. There were no other significant main effects or
interactions.
TABLE 3
Mean (and standard error) proportion priming for congruent targets in biasing
contexts relative to neutral baseline in Experiment 2 (12 dB SNR), based on inverse
efciency scores
Type of
competing signal
Channel of
presentation
(competing
signal)
Ear of
presentation
(context)
Presence of
competing signal
Bias
strength
Proportion
priming
Forward Same Left Isolation Strong 0.11 (0.029)
Weak 0.05 (0.023)
Competing speech Strong 0.011 (0.030)
Weak 0.017 (0.030)
Right Isolation Strong 0.10 (0.029)
Weak 0.05 (0.023)
Competing speech Strong 0.00 (0.030)
Weak 0.02 (0.030)
Different Left Isolation Strong 0.10 (0.030)
Weak 0.08 (0.025)
Competing speech Strong 0.11 (0.032)
Weak 0.01 (0.032)
Right Isolation Strong 0.13 (0.030)
Weak 0.08 (0.025)
Competing speech Strong 0.05 (0.032)
Weak 0.04 (0.032)
Time-reversed Same Left Isolation Strong 0.18 (0.030)
Weak 0.10 (0.025)
Competing speech Strong 0.01 (0.032)
Weak 0.04 (0.032)
Right Isolation Strong 0.07 (0.032)
Weak 0.02 (0.026)
Competing speech Strong 0.05 (0.034)
Weak 0.04 (0.034)
Different Left Isolation Strong 0.16 (0.030)
Weak 0.10 (0.025)
Competing speech Strong 0.12 (0.032)
Weak 0.02 (0.032)
Right Isolation Strong 0.15 (0.030)
Weak 0.08 (0.025)
Competing speech Strong 0.11 (0.032)
Weak 0.08 (0.032)
Note: Values less than zero reflect facilitation.
AUDITORY SENTENCE PRIMING IN COMPETING SPEECH 21
Downloaded by [King's College London] at 07:10 24 October 2011
Experiment 2B: Competing speech in a different channel
Four-way ANOVAs with bias strength and competing signal as within-
subjects variables and type of competing signal and ear of presentation as
TABLE 4
Mean (and standard error) proportion priming for incongruent targets in biasing
contexts relative to neutral baseline in Experiment 2 (12 dB SNR), based on inverse
efciency scores
Type of
competing signal
Channel of
presentation
(competing
signal)
Ear of
presentation
(context)
Presence of
competing signal
Bias
strength
Proportion
priming
Forward Same Left Isolation Strong 0.15 (0.070)
Weak 0.20 (0.073)
Competing speech Strong 0.01 (0.045)
Weak 0.00 (0.055)
Right Isolation Strong 0.16 (0.070)
Weak 0.16 (0.073)
Competing speech Strong 0.04 (0.045)
Weak 0.04 (0.055)
Different Left Isolation Strong 0.23 (0.073)
Weak 0.22 (0.077)
Competing speech Strong 0.15 (0.048)
Weak 0.16 (0.058)
Right Isolation Strong 0.16 (0.073)
Weak 0.18 (0.077)
Competing speech Strong 0.18 (0.048)
Weak 0.09 (0.058)
Time-reversed Same Left Isolation Strong 0.20 (0.073)
Weak 0.15 (0.077)
Competing speech Strong 0.08 (0.048)
Weak 0.03 (0.058)
Right Isolation Strong 0.24 (0.078)
Weak 0.25 (0.081)
Competing speech Strong 0.07 (0.051)
Weak 0.10 (0.061)
Different Left Isolation Strong 0.29 (0.073)
Weak 0.31 (0.077)
Competing speech Strong 0.19 (0.048)
Weak 0.31 (0.058)
Right Isolation Strong 0.10 (0.073)
Weak 0.13 (0.077)
Competing speech Strong 0.08 (0.048)
Weak 0.13 (0.058)
Note: Values greater than zero reflect inhibition.
22 AYDELOTT ET AL.
Downloaded by [King's College London] at 07:10 24 October 2011
between-subjects variables were conducted separately for congruent and
incongruent targets.
Congruent targets. Significant main effects emerged for both bias
strength, F(1, 32) 41.12, pB.0001, and competing signal, F(1, 32 8.58,
p B.01, such that facilitation effects were larger for strong contexts than
weak contexts, and for contexts presented in isolation than in competing
speech. In addition, a significant interaction of competing signal bias
0
0.05
0.1
0.15
0.2
0.25
0.3
Proportion Priming
-0.3
-0.25
-0.2
-0.15
-0.1
-0.05
0
Proportion Priming
isolation competing speech
**
**
weak biasstrong bias
weak biasstrong bias
isolation competing speech
Figure 4. Effect of competing speech presented in the same auditory channel as the sentence
context on the facilitation of congruent targets (top panel) and inhibition of incongruent targets
(bottom panel) in Experiment 2 (12 dB SNR).
AUDITORY SENTENCE PRIMING IN COMPETING SPEECH 23
Downloaded by [King's College London] at 07:10 24 October 2011
strengthear of presentation was obtained, F(1, 32) 12.43, pB.005. These
data are plotted in Figure 5 (top panel). Separate two-way repeated measures
ANOVAs were conducted for each ear of presentation (N18 in each group)
to clarify the nature of this interaction. As in Experiment 1, ear of
presentation effects are described as primarily reflecting processing in the
contralateral hemisphere.
–0.3
–0.25
–0.2
–0.15
–0.1
–0.05
0
strong bias
weak bias strong bias weak bias
LH Sentence RH Sentence
Proportion Priming
isolation competing speech
–0.3
–0.25
–0.2
–0.15
–0.1
–0.05
0
strong bias
weak bias strong bias weak bias
LH Sentence RH Sentence
Proportion Priming
isolation competing speech
ns
*
ns
ns
*
*
*
ns ns
*
ns
~
Figure 5. Effect of competing speech in a different auditory channel from the sentence context,
presented to the left hemisphere/right ear (left plots) or the right hemisphere/left ear (right plots)
on facilitation of congruent targets at SNRs of 12 dB in Experiment 2 (top panel), and 0 dB in
Experiment 1 (bottom panel).
24 AYDELOTT ET AL.
Downloaded by [King's College London] at 07:10 24 October 2011
Contexts presented to the right ear (left hemisphere). Results revealed a
significant main effect of bias strength, F(1, 16) 7.71, p B.05, with strong
contexts producing greater facilitation than weak contexts, and a marginally
significant competing signalbias strength interaction, F(1, 16) 3.46,
p.08. One-sample t-tests demonstrated significant priming for contexts
presented in isolation, strong: t(17) 6.36, p B.0001; weak: t(17)
4.12, pB.001, and in competing speech, strong: t(17) 5.15,
pB.0001; weak: t(17) 2.89, pB.01. However, the effect of bias
strength on facilitation emerged only for contexts presented in isolation,
paired t(17) 3.72, pB.005, and not in competing speech, paired
t(17) 1.28, p.22.
Contexts presented to the left ear (right hemisphere). Significant main
effects of both bias strength, F(1, 16) 55.23, pB.0001, and competing
signal, F(1, 16) 6.61, pB.05, emerged. Thus, greater facilitation emerged
for strong contexts than for weak contexts, and facilitation was reduced in
competing speech relative to isolation. In addition, there was a significant
competing signalbias strength interaction, F(1, 16) 9.12, pB.01. One-
sample t-tests revealed significant facilitation effects for contexts presented in
isolation, strong: t(17) 7.14, pB.0001; weak: t(17) 7.04, pB.0001.
However, for contexts presented in competing speech, significant facilitation
emerged only for strong contexts, t(17) 5.80, pB.0001, and not for weak
contexts, t(17) 0.33, p.75. Further, although competing speech had
a significant effect on facilitation for weak contexts, paired t(17) 4.05,
pB.001, it had no significant effect on strong contexts, paired t(17) 0.51,
p.62.
Comparison with results of Experiment 1. These results differed sub-
stantially from the pattern of facilitation obtained for competing speech in a
different auditory channel at an SNR of 0 dB (Experiment 1). This was
confirmed in a five-way ANOVA with bias strength and competing signal as
within-subjects variables and SNR, type of competing signal, and ear of
presentation as between-subjects variables, which revealed a significant
interaction of competing signalbias strengthear of presentation
SNR, F(1, 68) 9.78, pB.01. The data from the 0 dB SNR are plotted in
Figure 5 (bottom panel). Paired t-tests revealed that competing speech did
not significantly reduce facilitation at the 0 dB SNR when the sentence
context was presented to the RE/LH, strong contexts: t(19) 0.41,
p .69; weak contexts: t(19)0.32, p.75. The effect of bias strength on
facilitation was significant for sentences presented in isolation, t(19) 2.50,
pB.05, and in competing speech, t(19) 4.29, pB.0001. When the
sentence context was presented to the LE/RH, facilitation was significantly
reduced for strong contexts, t(19) 3.60, pB.01, but not for weak contexts,
AUDITORY SENTENCE PRIMING IN COMPETING SPEECH 25
Downloaded by [King's College London] at 07:10 24 October 2011
t(19) 1.42, p.17. The effect of bias strength on facilitation was significant
for sentences presented in isolation, t(19) 7.20, pB.0001, and marginally
significant in competing speech, t(19) 1.76, p.09.
Incongruent targets. There were no significant main effects. The inter-
action of bias strengthcompeting signal ear of presentation was margin-
ally significant, F(1, 32) 3.71, p.06. However, separate two-way
ANOVAs conducted for each ear of presentation revealed no significant
main effects of bias strength or competing signal and no significant
interactions. One-sample t-tests revealed that significant inhibition emerged
for sentence contexts presented in isolation, strong: t(35) 4.29, pB.0001;
weak: t(35) 4.60, p B.0001, and in competing speech, strong: t(35) 6.01,
pB.0001; weak: t(35) 4.88, pB.0001. These data are plotted in Figure 6.
Discussion
As in Experiment 1, biasing sentence contexts presented in isolation
produced both facilitation and inhibition priming, with significantly greater
facilitation emerging for strong contexts than weak contexts and no
difference in inhibition due to bias strength. Thus, the facilitation of
congruent targets was sensitive to both semantic expectancy and contextual
constraint, whereas the inhibition of incongruent targets was based on
expectancy alone and was not influenced by the number of plausible
completions suggested by the sentence context.
Competing speech presented in the same auditory channel at the more
demanding SNR of 12 dB eliminated facilitation priming for both strong
and weak contexts, and this effect emerged irrespective of the intelligibility of
the competing speech or the ear of presentation of the target sentence.
0
0.05
0.1
0.15
0.2
0.25
0.3
Proportion Priming
isolation competing speech
ns
ns
weak biasstrong bias
Figure 6. Effect of competing speech in a different auditory channel from the sentence context
on inhibition of incongruent targets in Experiment 2 (12 dB SNR).
26 AYDELOTT ET AL.
Downloaded by [King's College London] at 07:10 24 October 2011
However, significant inhibition priming was obtained for targets in strong
contexts at this SNR, indicating that the operation of task-specific processing
strategies that depend upon the output of the semantic system are relatively
robust to energetic masking. As strategic effects are slow to emerge (den
Heyer, Briand, & Smith, 1985; Neely, 1977, 1991), it is possible that these
processes benefit from later-stage perceptual reanalysis of the context signal
after initial sensory activation. Although no significant inhibition was
obtained for targets in weak contexts, the interaction of bias strength and
competing signal was not significant. Inhibition priming was not affected by
competing signal intelligibility or ear of presentation.
Competing speech presented in a separate auditory channel from the
sentence context produced a strikingly different pattern of facilitation at the
12 dB SNR than was observed at the more moderate SNR of 0 dB in
Experiment 1. Firstly, whereas the interference produced by the competing
signal in Experiment 1 depended upon its intelligibility, in Experiment 2 both
forward and time-reversed speech disrupted facilitation. This suggests that
the extent to which the meaningful content of the competing signal increases
informational masking depends upon the perceptual demands of the
listening environment, with the intelligibility of the competing signal
contributing less to the overall masking effect under more challenging
conditions.
As predicted, the reduction in SNR appears also to have increased
listeners focus of attention on the target signal, enhancing the processing of
the context sentence while suppressing the processing of the competing
speech. This would account at the counterintuitive finding that facilitation of
targets in strong contexts presented to the LE/RH was significantly reduced
at the 0 dB SNR, but was unaffected at the 12 dB SNR. Whereas the weak
focus of attention at the moderate SNR gave rise to a REA for intelligible
competing speech in this condition, the stronger focus at the more
demanding SNR may have served to override the REA (cf. Hugdahl,
2005), eliminating the interference effect for strong contexts. However, weak
contexts presented to the LE/RH failed to produce significant facilitation at
the 12 dB SNR. This raises the possibility that expectancy-based
facilitation does not emerge when contextual information is presented
exclusively to the LE/RH while processing of RE/LH stimuli is suppressed.
According to this interpretation, the pattern of performance observed in this
condition reflects message-level semantic processing in the RH only.
Interestingly, the results suggest that facilitation in the RH is based solely
on contextual constraint with no influence of expectancy. Similarly, in the
present study, when the context is presented to the RE/LH with competing
speech presented to the LE/RH at this SNR, significant facilitation emerges
for targets in both strong and weak contexts with no significant difference
AUDITORY SENTENCE PRIMING IN COMPETING SPEECH 27
Downloaded by [King's College London] at 07:10 24 October 2011
between these conditions. This suggests that facilitation in the LH is
expectancy-based with no influence of contextual constraint.
A similar pattern of results was observed in a recent neurophysiological
study using event-related brain potentials (ERPs) to investigate semantic
processing of words in context in the visual modality. The N400 response*a
negative deflection in the ERP signal maximal over centro-parietal electrode
sites observed approximately 400 ms after the presentation of a word*is
sensitive to the sentence context in which a word occurs, such that N400
amplitude is reduced for words that are congruent with the sentence context
and enhanced for incongruent words (for a review see Kutas & Federmeier,
2000). Wlotko and Federmeier (2007) compared the effects of semantic
expectancy and contextual constraint on the N400 response to visual words
presented to the right visual field/LH and the left visual field/RH. Sentence
contexts were either strongly constraining (with few plausible completions)
or weakly constraining (with many plausible completions), and targets were
congruent words that were either expected or unexpected on the basis of the
context as determined by cloze probability measures. A reduction in N400
amplitude for expected targets was observed in both strongly and weakly
constraining contexts for targets presented to the RVF/LH. In contrast,
expected targets presented to the LVF/RH showed a large reduction in N400
amplitude in strongly constraining contexts but no reduction in weakly
constraining contexts. Thus, the N400 for RVF/LH targets was sensitive to
target expectancy, independent of contextual constraint, whereas for LVF/
RH targets, the N400 was sensitive to constraint, independent of expectancy.
Wlotko and Federmeier interpret these results in terms of the distinct roles of
prediction and constraint in the processing of words in a semantic context.
According to this view, the LH uses expectancies based on the semantic
features activated by the sentence context to predict upcoming information.
Targets presented to the RVF benefit when these expectancies are confirmed,
irrespective of the number of plausible completions the semantic context
allows. In contrast, the RH is responsible for integrating words into the
ongoing semantic context. Thus, an advantage is observed for an expected
word presented to the LVF when there are fewer plausible completions and no
such advantage when there are many plausible completions. The pattern of
facilitation effects observed in the present study is consistent with this account,
and suggests that similar results may be obtained in the auditory modality
under dichotic listening conditions using contralateral speech masking.
Unlike the results for facilitation, the inhibition of incongruent targets was
not significantly affected by competing speech in a separate auditory
channel, even at the more demanding SNR. Thus, contrary to our original
predictions, inhibitory priming is preserved under highly challenging
listening conditions when the attended and competing signals are presented
28 AYDELOTT ET AL.
Downloaded by [King's College London] at 07:10 24 October 2011
at different spatial locations. This finding indicates that increased demands
on selective auditory attention in a multitalker environment do not interfere
with the use of semantic information in the generation of task-specific
processing strategies.
GENERAL DISCUSSION
This study investigated the effect of a single competing talker on the priming
of spoken words by a meaningful sentence context. The sentence priming
paradigm was selected as an online measure of auditory comprehension that
is sensitive to both the activation of the semantic system, as reflected
primarily in the facilitation of congruent targets, and the operation of later-
stage strategic processes that use semantic information to aid task
performance, as reflected primarily in the inhibition of incongruent targets.
The experimental paradigm explored the role of the following factors in
sentence comprehension in competing speech: (1) the strength of the
semantic bias imposed by the sentence context, as a means of evaluating
the effects of semantic expectancy and contextual constraint on spoken word
recognition in quiet and speech-masked conditions; (2) the energetic masking
effect of competing speech and the associated benefit from binaural spatial
unmasking, by comparing performance when the competing signal was
presented in the same auditory channel as the sentence context and when the
competing signal was presented in a separate auditory channel; (3)
the contribution of meaningful content to the informational masking effect
of competing speech, by comparing the effects of forward and time-reversed
competing signals; (4) the perceptual and attentional demands of the
listening situation, by presenting the attended and competing signals at
moderate and more challenging SNRs; and (5) hemispheric differences in
speech perception and semantic processing, by manipulating the ear of
presentation of the attended and competing signals.
Overall, the findings supported previous claims that facilitation and
inhibition effects reflect distinct aspects of processing in primed lexical
decision (Aydelott & Bates, 2004; Neely, 1991; Stanovich & West, 1983). The
facilitatory and inhibitory components of the priming effect responded
differently to the experimental manipulations under investigation, and the
pattern of results was compatible with the proposal that facilitation reflects
online semantic activation and integration, whereas inhibition reflects the
operation of expectancy-based strategies specific to the lexical decision task.
For sentence contexts presented in isolation, responses to congruent targets
were faster and more accurate in both the strong and weak bias conditions,
consistent with priming models in which the context generates broad
expectancies based the set of activated features in semantic memory, and
AUDITORY SENTENCE PRIMING IN COMPETING SPEECH 29
Downloaded by [King's College London] at 07:10 24 October 2011
words with compatible features are more efficiently processed (Schwanen-
flugel & LaCount, 1988). Responses to congruent targets were also faster and
more accurate in the strong bias than in the weak bias conditions, consistent
with models in which semantic integration is more easily achieved when a
sentence has fewer plausible completions, resulting in an additional boost to
facilitation when the context is highly constraining (Kutas & Federmeier,
2000; Traxler & Foss, 2000). In contrast, responses to incongruent targets
were slower and less accurate in both strong and weak contexts, but there was
no significant difference between these bias conditions. This pattern is
predicted under strategic accounts of inhibitory priming in lexical decision,
in which the violation of semantic expectancies and/or the absence of a
semantic match between the context and target triggers a ‘‘nonword’’
response that must be suppressed when the target is a real word (Becker,
1980, 1982; Neely, 1991; Neely et al., 1989; Stanovich & West, 1983).
According to this view, any semantic mismatch will produce inhibition,
irrespective of the strength of the contextual bias.
Facilitation and inhibition effects also differed in terms of their vulner-
ability to competing speech. Facilitation priming was reduced or eliminated
by a competing signal presented in the same auditory channel as the sentence
context, with weak contexts more adversely affected than strong contexts at
the moderate SNR. Thus, as observed in previous studies, the activation of
the semantic system is sensitive to factors such as energetic masking that
affect the intelligibility of the sensory input (Aydelott & Bates, 2004);
however, the extent to which reduced intelligibility disrupts word recognition
in context also depends upon the degree of semantic constraint imposed by
the preceding sentence (cf. Pichora-Fuller, 2008).
Facilitation priming showed a substantial overall benefit from binaural
unmasking when competing speech was presented in a different ear from the
sentence context. However, significant effects of ear of presentation revealed
hemispheric differences in both the interference produced by the competing
signal and the facilitation generated by the biasing sentence. When the
sentence and competing speech were presented at the same intensity level,
intelligible competing speech presented to the RE/LH interfered with the
processing of sentences presented to the LE/RH, resulting in reduced
facilitation of congruent targets in both strong and weak contexts. However,
no such interference was observed for time-reversed competing speech or for
forward competing speech presented to the LE/RH. Thus, at a moderate SNR
of 0 dB, facilitation priming appears to reflect the REA for intelligible speech
observed in previous studies (Kimura, 1961; Studdert-Kennedy & Shankwei-
ler, 1970; Wada & Rasmussen, 1960). In contrast, competing speech presented
at a greater intensity level than the context served to focus attention on the
auditory channel containing the biasing sentence, suppressing the processing
30 AYDELOTT ET AL.
Downloaded by [King's College London] at 07:10 24 October 2011
of competing speech in the ignored channel and eliminating the effect of
competing signal intelligibility. Further, enhanced attention to a single
auditory channel at the 12 dB SNR revealed different patterns of priming
depending upon the ear of presentation of the context, with sentences
presented to the RE/LH producing facilitation of targets in both strong and
weak contexts, and sentences presented to the LE/RH producing facilitation
of targets in strong contexts only. This finding supports the conclusions of
Wlotko and Federmeier (2007) that the LH generates predictions about
upcoming information based on the semantic features activated by the
context, whereas the RH integrates incoming lexical-semantic information
into the sentence-level meaning. Federmeier (2007) accounts for this hemi-
spheric asymmetry in semantic processing in terms of the role of the RH in the
bottom-up assimilation of newly presented linguistic information, and the
potential contribution of left-lateralised language production mechanisms to
the generation of semantic predictions (cf. Pickering & Garrod, 2007).
Whether the language production system plays a critical role in predicting
upcoming information in spoken language comprehension remains a topic for
future research.
Unlike facilitation priming, the inhibition of incongruent targets was not
reliably influenced by the intelligibility of the competing signal or the ear of
presentation of the sentence context at either SNR. Competing speech in the
same auditory channel reduced inhibition priming at both SNRs; however,
significant inhibition was obtained for weak contexts at the 0 dB SNR, and
for strong contexts at the 12 dB SNR*conditions under which facilitation
priming was eliminated. This suggests that the generation of expectancy-
based strategies is relatively robust to reductions in signal intelligibility
introduced by energetic masking. The slow time course of strategic
processing may allow top-down perceptual reanalysis of the degraded signal,
making additional semantic information available for expectancy generation
at a later stage. Inhibition priming was unaffected when competing speech
was presented in a separate auditory channel from the sentence context,
demonstrating a complete release from masking at both SNRs. Thus,
although inhibitory priming effects are vulnerable to reductions in processing
time, as in the case of time-compressed speech (Aydelott & Bates, 2004),
task-dependent strategies are relatively insensitive to increased demands on
auditory selective attention.
In conclusion, the results of the present study supported the prediction
that the interference produced by a competing speech signal on the
processing of words in a semantic context depends upon the meaningful
content of the competing signal, the attentional demands of the listening
environment, and hemispheric asymmetries in the processing of speech and
semantic information. The energetic masking effect of a competing speech
AUDITORY SENTENCE PRIMING IN COMPETING SPEECH 31
Downloaded by [King's College London] at 07:10 24 October 2011
signal presented in the same auditory channel as an attended sentence
context reduced or eliminated the facilitation of congruent targets, particu-
larly for weakly biasing sentences, whereas the inhibition of incongruent
targets persisted even at a demanding SNR of 12 dB. Competing speech
presented to a different auditory channel from the attended signal produced
significant ear-of-presentation effects on facilitation priming, revealing
hemispheric differences in the processing of both the competing and attended
signals. At a 0 dB SNR, meaningful competing speech interfered with
facilitation only when presented to the right ear, consistent with the REA for
intelligible speech. At a 12 dB SNR, however, an interaction of ear of
presentation and contextual strength was observed, indicating that the extent
to which facilitation effects reflect semantic expectancy or contextual
constraint differs according to whether the sentence context is presented to
the RE/LH or the LE/RH. The findings were consistent with previous ERP
studies using visually presented materials. The dichotic sentence priming
paradigm therefore offers a potential methodology for exploring hemispheric
differences in semantic processing in the auditory modality.
Manuscript received 4 October 2010
Revised manuscript received 15 May 2011
First published online XXXX
REFERENCES
Arons, B. (1992). A review of the cocktail party effect. Journal of the American Voice I/O Society,
12(July), 3550.
Aydelott, J., & Bates, E. (2004). Effects of acoustic distortion and semantic context on lexical
access. Language and Cognitive Processes, 19(1), 2956.
Becker, C. A. (1980). Semantic context effects in visual word recognition: An analysis of semantic
strategies. Memory & Cognition, 8(6), 493512.
Becker, C. A. (1982). The development of semantic context effects: Two processes or two strategies?
Reading Research Quarterly, 17(4), 482502.
Bilger, R. C., Nuetzel, J. M., Rabinowitz, W. M., & Rzeczkowski, C. (1984). Standardization of a
test of speech perception in noise. Journal of Speech and Hearing Research, 27(1), 3248.
Boersma, P. (2001). Praat, a system for doing phonetics by computer. Glot International, 5(9/10),
341345.
Boulenger, V., Hoen, M., Ferragne, E., Pellegrino, F., & Meunier, F. (2010). Real-time lexical
competitions during speech-in-speech comprehension. Speech Communication, 52(3), 246253.
Bronkhorst, A. W. (2000). The cocktail party phenomenon: A review of research on speech
intelligibility in multiple-talker conditions. Acta Acustica, 86, 117128.
Brown, G. D. (1984). A frequency count of 190,000 words in the London-Lund Corpus of English
Conversation. Behavior Research Methods, Instruments, and Computers, 6, 502532.
Brungart, D. S. (2001). Informational and energetic masking effects in the perception of
two simultaneous talkers. The Journal of the Acoustical Society of America, 109(3), 11011109.
Burke, D. M., & Yee, P. L. (1984). Semantic priming during sentence processing by young and older
adults. Developmental Psychology, 20(5), 903910.
32 AYDELOTT ET AL.
Downloaded by [King's College London] at 07:10 24 October 2011
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), 975979.
Coltheart, M. (1981). The MRC Psycholinguistic Database. Quarterly Journal of Experimental
Psychology, 33A, 495505.
den Heyer, K., Briand, K., & Smith, L. (1985). Automatic and strategic effects in semantic
priming: An examination of Beckers verication model. Memory & Cognition, 13(3), 228232.
Federmeier, K. D. (2007). Thinking ahead: The role and roots of prediction in language
comprehension. Psychophysiology, 44(4), 491505.
Fischler, I., & Bloom, P. (1979). Automatic and attentional processes in the effects of
sentence contexts on word recognition. Journal of Verbal Learning and Verbal Behavior, 18(1),
120.
Fischler, I., & Bloom, P. A. (1980). Rapid processing of the meaning of sentences. Memory &
Cognition, 8(3), 216225.
Freyman, R., Balakrishnan, U., & Helfer, K. S. (2001). Spatial release from informational
masking in speech recognition. The Journal of the Acoustical Society of America, 109,
21122122.
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),
24452454.
Hugdahl, K. (2005). Symmetry and asymmetry in the human brain. European Review, 13(S2),
119133.
Hutcherson, R. W., Dirks, D. D., & Morgan, D. E. (1979). Evaluation of the speech perception in
noise (SPIN) test. Otolaryngology and Head and Neck Surgery, 87(2), 239245.
Kalikow, D. N., Stevens, K. N., & Elliott, L. L. (1977). Development of a test of speech
intelligibility in noise using sentence materials with controlled word predictability. The Journal
of the Acoustical Society of America, 61(5), 13371351.
Kimura, D. (1961). Cerebral dominance and the perception of verbal stimuli. Canadian Journal of
Psychology, 15(3), 166171.
Kucera, H., & Francis, W. N. (1967). Computational analysis of present-day American English.
Providence, RI: Brown University Press.
Kutas, M., & Federmeier, K. D. (2000). Electrophysiology reveals semantic memory use in language
comprehension. Trends in Cognitive Sciences, 4(12), 463470.
McNamara, T. P. (2005). Semantic priming: Perspectives from memory and word recognition.Hove,
UK: Psychology Press.
Meyer, D. E., & Schvaneveldt, R. W. (1971). Facilitation in recognizing pairs of words: Evidence
of a dependence between retrieval operations. JournalofExperimentalPsychology, 90(2), 227234.
Neely, J. H. (1977). Semantic priming and retrieval from lexical memory: Roles of inhibitionless
spreading activation and limited-capacity attention. Journal of Experimental Psychology:
General, 106(3), 226254.
Neely, J. H. (1991). Semantic priming effects in visual word recognition: A selective review of
current ndings and theories. In D. Besner & G. W. Humphreys (Eds.), Basic processes in
reading: Visual word recognition (pp. 264336). Hillsdale, NJ: Lawrence Erlbaum.
Neely, J. H., & Keefe, D. E. (1989). Semantic context effects on visual word processing: A
hybrid prospective/retrospective processing theory. In G. H. Bower (Ed.), The psychology of
learning and motivation: Advances in research and theory (pp. 207248). San Diego, CA:
Academic Press.
Neely, J. H., Keefe, D. E., & Ross, K. L. (1989). Semantic priming in the lexical decision task: Roles
of prospective prime-generated expectancies and retrospective semantic matching. Journal of
Experimental Psychology: Learning, Memory, and Cognition, 15(6), 10031019.
AUDITORY SENTENCE PRIMING IN COMPETING SPEECH 33
Downloaded by [King's College London] at 07:10 24 October 2011
Pichora-Fuller, M. K. (2008). Use of supportive context by younger and older adult listeners:
Balancing bottom-up and top-down information processing. International Journal of Audiology,
47(S2), S72S82.
Pickering, M. J., & Garrod, S. (2007). Do people use language production to make predictions
during comprehension? Trends in Cognitive Sciences, 11(3), 105110.
Plaut, D. C., & Booth, J. R. (2000). Individual and developmental difference: Empirical and
computational support for a single-mechanism account of lexical processing. Psychological
Review, 107(4), 786823.
Plaut, D. C., & Booth, J. R. (2006). More modeling but still no stages: Reply to Borowsky and
Besner. Psychological Review, 113(1), 196200.
Ratcliff, R. (1993). Methods for dealing with reaction time outliers. Psychological Bulletin, 114(3),
510532.
Ratcliff, R., & McKoon, G. (1981). Automatic and strategic priming in recognition. Journal of
Verbal Learning and Verbal Behavior, 20(2), 204215.
Schuberth, R. E., & Eimas, P. D. (1977). Effects of context on the classication of words
and nonwords. Journal of Experimental Psychology: Human Perception and Performance, 3(1), 2736.
Schwanenugel, P. J., & LaCount, K. L. (1988). Semantic relatedness and the scope of facilitation
for upcoming words in sentences. Journal of Experimental Psychology: Learning, Memory, and
Cognition, 14(2), 344354.
Schwanenugel, P. J., & Shoben, E. J. (1985). The inuence of sentence constraint on the scope of
facilitation for upcoming words. Journal of Memory and Language, 24(2), 232252.
Schwanenugel, P. J., & White, C. R. (1991). The inuence of paragraph information on the
processing of upcoming words. Reading Research Quarterly, 26(2), 160177.
Scott, S. K., Blank, C. C., Rosen, S., & Wise, R. J. (2000). Identication of a pathway for intelligible
speech in the left temporal lobe. Brain, 123, 24002406.
Slywotzky, A. J. (1999). Profit patterns: Patterns: 30 ways to anticipate and profit from strategic
forces. Hoboken, NJ: Wiley.
Stanovich, K. E., & West, R. F. (1983). On priming by a sentence context. Journal of Experimental
Psychology: General, 112(1), 136.
Studdert-Kennedy, M., & Shankweiler, D. (1970). Hemispheric specialization for speech percep-
tion. Journal of the Acoustical Society of America, 48(2), 579594.
Townsend, J. T., & Ashby, F. G. (1983). The stochastic modeling of elementary psychological
processes. Cambridge, UK: Cambridge University Press.
Traxler, M. J., & Foss, D. J. (2000). Effects of sentence constraint on priming in natural language
comprehension. Journal of Experimental Psychology: Learning, Memory, and Cognition, 26(5),
12661282.
Tyler, L. K., Randall, B., & Marslen-Wilson, W. D. (2002). Phonology and neuropsychology of the
English past tense. Neuropsychologia, 40(8), 11541166.
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), 519526.
Wada, J., & Rasmussen, T. (1960). Intracarotid injection of sodium amytal for the lateralization of
cerebral speech dominance. Journal of Neurosurgery, 17(2), 266282.
West, R. F., & Stanovich, K. E. (1982). Source of inhibition in experiments on the effect of sentence
context on word recognition. Journal of Experimental Psychology: Learning, Memory, and
Cognition, 8(5), 385399.
Wlotko, E. W., & Federmeier, K. D. (2007). Finding the right word: Hemispheric asymmetries in
the use of sentence context information. Neuropsychologia, 45(13), 30013014.
Wood, N. L., & Cowan, N. (1995). The cocktail party phenomenon revisited: Attention and
memory in the classic selective listening procedure of Cherry (1953). Journal of Experimental
Psychology: General, 124(3), 243262.
34 AYDELOTT ET AL.
Downloaded by [King's College London] at 07:10 24 October 2011
APPENDIX
Sentence contexts and word targets
Strong bias context Weak bias context
Congruent
target
One day the prince will become the One of the important pieces in chess is
the
king
He paid the rent on the first day of the They took a long holiday and stayed
away for almost a
month
While they waited for their table, they
had a drink at the
The maitre dinvited them to sit at the bar
Pasta is my favourite kind of Before making dinner, I had to buy the food
The boy bounced the The man threw the ball
The woman had been ill so often she
began to worry about her
They drank to his health
The jockey went over the fence on his The rancher called out the vet to see
his
horse
At the hotel, we went swimming in the It made a big splash when I pushed my
brother into the
pool
To cross the river, they had to go over
the
The construction workers built a bridge
He weighed himself on the At the post office, I put the letters on
the
scale
The boy went to the library to borrow a The nervous student reached over to
the shelf to get the
book
When her husband left her, it broke her The surgeon transplanted the patients heart
The pendant hung on the end of the
silver
At the jewellery shop, she bought a
pretty
chain
Before receiving a licence, every driver
must pass a
He was very nervous about the test
I ran upstairs to answer the When we forgot to pay our bill, they
cut off our
phone
She always writes with her left The child was playing with matches
and burnt his
hand
At the wedding, they admired the
brides white
In the clothes shop, she bought a shirt
and a
dress
She got on all fours to scrub the The woman spilt her drink on the floor
He asked her out on a He apologised to her for forgetting
their
date
If he missed one more day at work, hed
lose his
She didnt want to go to work because
she hated her
job
In the stuffy room, it was difficult for
her to take a
The child blew out all of the candles in
a single
breath
When she woke up, she realised it had
all been a
The sleeping princess was so beautiful,
she was like a
dream
Granddad always sits in his favourite At the furniture shop, I bought a chair
The day he was elected, the politician
gave a
The politician was very nervous about
the
speech
There are seven days in a She stayed in London for a week
The runner finished third in the I hoped the competitor would win the race
AUDITORY SENTENCE PRIMING IN COMPETING SPEECH 35
Downloaded by [King's College London] at 07:10 24 October 2011
Appendix (Continued )
Strong bias context Weak bias context
Congruent
target
Before he coughed, he felt a tickle in his The hot tea was good for her throat
The vicar declared they were husband
and
He bought a birthday present for his wife
She took the message and wrote her
flatmate a
They werent at home, so she wrote
them a
note
A rectangle is a common The geometry student drew the shape
I tell all of my secrets to my closest When I go to a party I sometimes
bring a
friend
The boy had a cold and needed to blow
his
The clown wore makeup on his nose
She moved to the front to get a better The tourists admired the beautiful view
The astronauts landed on the That night, I looked up at the moon
A slate fell off the The builder came and patched the roof
She was still grieving a year after her
mothers
The family was distraught over his death
The tour guide preferred to lead a small The students worked together in a group
They went rowing in a As the sailor looked out at the sea he
saw a
boat
In spite of her strict diet she couldnt
lose the
The railway platform couldnt
withstand the
weight
Children learn most when theyre at a
good
When I was a child I hated everyone in
my
school
The man with the crutches had broken
his
The nurse bandaged the patients leg
The grizzly cub would soon mature into
an adult
In the forest, they saw a bear
She couldnt stand to see the look on his The baby had porridge all over his face
In the office, the computer is on my They took all of the furniture in the
office except for the
desk
The giraffe has a long The acrobat had broken his neck
He went to the cellar to fetch the white Before the drinks party, he went to the
off-licence to buy the
wine
She hung the painting on the Mary leaned against the wall
He wanted to come in, but she refused
to open the
Her husband told her to open the door
She sat on a bench and fed the pigeons
in the
We went for a walk in the park
For Valentines Day he gave her a red The gardeners favourite flower was a rose
The baby had a dummy in its The chocolate melted in his mouth
I went to the station to catch the At 12:35 she got on the train
It was getting dark, so she turned on
the
At night, he couldnt work without a light
The barristers objection was overruled
by the
The defendants witness testified
before the
judge
The centre forward scored his third The player argued with the referee
about the
goal
36 AYDELOTT ET AL.
Downloaded by [King's College London] at 07:10 24 October 2011
Appendix (Continued )
Strong bias context Weak bias context
Congruent
target
The criminal knew that he was breaking
the
The solicitor always obeyed the law
She was afraid to fly and refused to get
on the
She liked to travel on a plane
The businessmen signed the papers
after closing the
The car salesman convinced them it
was a good
deal
They went to the cinema to see a Last Saturday on television I watched
a
film
Theres lots of sand on the In the summer we sometimes go to the beach
AUDITORY SENTENCE PRIMING IN COMPETING SPEECH 37
Downloaded by [King's College London] at 07:10 24 October 2011
... The facilitatory effect of semantic context on the activation of words in native listening is however highly sensitive to the intelligibility of the speech, i.e., the semantic context produces less activation of semantically related items in degraded listening conditions (Aydelott & Bates, 2004). Nevertheless, semantic information facilitates word recognition in degraded listening conditions for native listeners (Aydelott & Bates, 2004;Aydelott et al., 2012). ...
... Primes thus have a direct effect on the multiple activation process, and on speed of recognition. This priming effect is smaller in non-native compared to native listening in noise (Aydelott et al., 2012;Golestani et al., 2009). Second, there is an effect of predictability (or semantic congruency) above and beyond priming which has been related to the fit of the activated word in the sentence context (referred to as semantic integration) (FitzPatrick & Indefrey, 2010;Zwitserlood, 1989). ...
... Moreover, differences in exposure to the language would also explain the differences between native and non-native listeners in the ability to use semantic contextual information (e.g., Aydelott & Bates, 2004;Aydelott et al., 2012;Bradow & Alexander, 2007;Golestani et al., 2009;Mayo et al., 1997), and would be in line with findings by Ezzatian et al. (2010) and Cooke et al. (2008), who found that a listener's ability (to pick up and use specific acoustic cues) to separate background noise from target speech is a more important factor in explaining why some people suffer less from the presence of background noise than other listeners than the listener's proficiency in the non-native language. We thus argue that native and non-native listening can and should thus be viewed as lying on a continuum of language experience, with a corresponding effect of noise on spoken-word recognition. ...
Article
Full-text available
There is ample evidence that recognising words in a non-native language is more difficult than in a native language, even for those with a high proficiency in the non-native language involved, and particularly in the presence of background noise. Why is this the case? To answer this question, this paper provides a systematic review of the literature on non-native spoken-word recognition in the presence of background noise, and posits an updated theory on the effect of background noise on native and non-native spoken-word recognition. The picture that arises is that although spoken-word recognition in the presence of background noise is harder in a non-native language than in one's native language, this difference is not due to a differential effect of background noise on native and non-native listening. Rather, it can be explained by differences in language exposure, which influences the uptake and use of phonetic and contextual information in the speech signal for spoken-word recognition.
... . These studies showed that the gap in accuracy between native and non-native listeners increases as the listening conditions become more challenging. Some studies evaluate the effect of semantic context on target word recognition [14]. Some of these methods have studied the effect of listener's proficiency [1]. ...
... These studies showed that the gap in accuracy between native and non-native listeners increases as the listening conditions become more challenging. Some studies evaluate the effect of semantic context on target word recognition [14]. Some of these methods have studied the effect of listener's proficiency [1]. ...
Preprint
In this study, listeners of varied Indian nativities are asked to listen and recognize TIMIT utterances spoken by American speakers. We have three kinds of responses from each listener while they recognize an utterance: 1. Sentence difficulty ratings, 2. Speaker difficulty ratings, and 3. Transcription of the utterance. From these transcriptions, word error rate (WER) is calculated and used as a metric to evaluate the similarity between the recognized and the original sentences.The sentences selected in this study are categorized into three groups: Easy, Medium and Hard, based on the frequency ofoccurrence of the words in them. We observe that the sentence, speaker difficulty ratings and the WERs increase from easy to hard categories of sentences. We also compare the human speech recognition performance with that using three automatic speech recognition (ASR) under following three combinations of acoustic model (AM) and language model(LM): ASR1) AM trained with recordings from speakers of Indian origin and LM built on TIMIT text, ASR2) AM using recordings from native American speakers and LM built ontext from LIBRI speech corpus, and ASR3) AM using recordings from native American speakers and LM build on LIBRI speech and TIMIT text. We observe that HSR performance is similar to that of ASR1 whereas ASR3 achieves the best performance. Speaker nativity wise analysis shows that utterances from speakers of some nativity are more difficult to recognize by Indian listeners compared to few other nativities
... Smaller N400s are found when semantic judgements are easier -a pair of congruent stimuli will elicit a smaller or no N400 in comparison to a pair of incongruent stimuli, because semantic analysis and access is easier in the first instance. The magnitude of the N400 is also reduced under degraded listening conditions (Aydelott, Baer-Henney et al., 2012;Carey, Mercure et al., 2014;Strauß, Kotz et al., 2013), an effect thought to reflect the reduced quality of the information reaching the semantic system (Aydelott, Dick et al., 2006). ...
... N400 responses were significantly reduced under degraded listening conditions, indicating semantic processes were affected by quality of speech perception. This finding replicates previous N400 studies with neurotypical populations (Aydelott, Baer-Henney et al., 2012;Carey, Mercure et al., 2014;Strauß, Kotz et al., 2013) and supports the hypothesis that the reduced N400 responses found in WA group may be an indirect effect of a phonological perception impairment. A further explanation of reduced N400 responses in WA implicates poor lexical access. ...
Article
Full-text available
Comprehension impairments in Wernicke's aphasia are thought to result from a combination of impaired phonological and semantic processes. However, the relationship between these cognitive processes and language comprehension has only been inferred through offline neuropsychological tasks. This study used ERPs to investigate phonological and semantic processing during online single word comprehension. EEG was recorded in a group of Wernicke's aphasia n=8 and control participants n=10 while performing a word-picture verification task. The N400 and Phonological Mapping Negativity/ Phonological Mismatch Negativity (PMN) event-related potential components were investigated as an index of semantic and phonological processing, respectively. Individuals with Wernicke's aphasia displayed reduced and inconsistent N400 and PMN effects in comparison to control participants. Reduced N400 effects in the WA group were simulated in the control group by artificially degrading speech perception. Correlation analyses in the Wernicke's aphasia group found that PMN but not N400 amplitude was associated with behavioural word-picture verification performance. The results confirm impairments at both phonological and semantic stages of comprehension in Wernicke's aphasia. However, reduced N400 responses in Wernicke's aphasia are at least partially attributable to earlier phonological processing impairments. The results provide further support for the traditional model of Wernicke's aphasia which claims a causative link between phonological processing and language comprehension impairments.
... Integration of redundant cues from multiple senses, such as vision and touch (Helbig & Ernst, 2007) may also depend on the reliability of those particular senses. For the purposes of this article we will refer to the phenomenon by which multiple cues index the same category feature as redundancy (although biological systems theorists would call this degeneracy), and N.B. there are compensatory strategies besides redundancy/degeneracy that may assist with perceptual impairments as well, such as those involving top-down predictive processes (e.g., Aydelott, Baer-Henney, Trzaskowski, Leech, & Dick, 2012;Ganong, 1980;Wicha, Moreno, & Kutas, 2004). ...
Article
Perception involves integration of multiple dimensions that often serve overlapping, redundant functions, for example, pitch, duration, and amplitude in speech. Individuals tend to prioritize these dimensions differently (stable, individualized perceptual strategies), but the reason for this has remained unclear. Here we show that perceptual strategies relate to perceptual abilities. In a speech cue weighting experiment (trial N = 990), we first demonstrate that individuals with a severe deficit for pitch perception (congenital amusics; N = 11) categorize linguistic stimuli similarly to controls (N = 11) when the main distinguishing cue is duration, which they perceive normally. In contrast, in a prosodic task where pitch cues are the main distinguishing factor, we show that amusics place less importance on pitch and instead rely more on duration cues-even when pitch differences in the stimuli are large enough for amusics to discern. In a second experiment testing musical and prosodic phrase interpretation (N = 16 amusics; 15 controls), we found that relying on duration allowed amusics to overcome their pitch deficits to perceive speech and music successfully. We conclude that auditory signals, because of their redundant nature, are robust to impairments for specific dimensions, and that optimal speech and music perception strategies depend not only on invariant acoustic dimensions (the physical signal), but on perceptual dimensions whose precision varies across individuals. Computational models of speech perception (indeed, all types of perception involving redundant cues e.g., vision and touch) should therefore aim to account for the precision of perceptual dimensions and characterize individuals as well as groups. (PsycINFO Database Record (c) 2019 APA, all rights reserved).
... (although biological systems theorists would call this 'degeneracy'), and N.B. there are compensatory strategies besides redundancy/degeneracy that may assist with perceptual impairments as well, such as those involving 'top-down' predictive processes (e.g. Ganong, 1980;Aydelott et al., 2012;Wicha et al., 2004). ...
Article
Perception involves integration of multiple dimensions that often serve overlapping, redundant functions, e.g. pitch, duration, and amplitude in speech. Individuals tend to prioritize these dimensions differently (stable, individualized perceptual ‘strategies’) but the reason for this has remained unclear. Here we show that perceptual strategies relate to perceptual abilities. In a speech cue weighting experiment (trial N = 990), we first demonstrate that individuals with a severe deficit for pitch perception (congenital amusics; N=11) categorize linguistic stimuli similarly to controls (N=11) when the main distinguishing cue is duration, which they perceive normally. In contrast, in a prosodic task where pitch cues are the main distinguishing factor, we show that amusics place less importance on pitch and instead rely more on duration cues—even when pitch differences in the stimuli were large enough for amusics to discern. In a second experiment testing musical and prosodic phrase interpretation (N=16 amusics; 15 controls), we found that relying on duration allowed amusics to overcome their pitch deficits to perceive speech and music successfully. We conclude that auditory signals, because of their redundant nature, are robust to impairments for specific dimensions, and that optimal speech and music perception strategies depend not only on invariant acoustic dimensions (the physical signal), but on perceptual dimensions whose precision varies across individuals. Computational models of speech perception (indeed, all types of perception involving redundant cues e.g. vision and touch) should therefore aim to account for the precision of perceptual dimensions and characterize individuals as well as groups.
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
Subliminally presented words have been shown to cause priming at orthographic and semantic levels. Here, we investigate whether subliminal priming can also occur at the syntactic level, and use such priming as a tool to probe the architecture for processing the syntactic features of written words. We studied the impact of masked and unmasked written word primes on response times to a subsequent visible target that shared or did not share syntactic features such as grammatical category and grammatical number. Methodological precautions included the use of distinct lists of subliminal primes that were never consciously seen, and the verification that participants were at chance in a prime-classification task. Across five experiments, subliminal priming could be induced by the repetition of the same grammatical category (e.g. a noun followed by another noun), by the transition between two categories (e.g. a determiner followed by a noun), or by the repetition of a single grammatical feature, even if syntax is violated (e.g. "they lemons", where the expression is ungrammatical but the plural feature is repeated). The orthographic endings of prime words also provided unconscious cues to their grammatical category. Those results indicate the existence of a representation of abstract syntactic features, shared between several categories of words, and which is quickly and unconsciously extracted from a flashed visual word.
Conference Paper
Speech-based information is usually communicated to users in a sequential manner, but users are capable of obtaining information from multiple voices concurrently. This fact implies that the sequential approach is possibly under-utilizing human perception capabilities to some extent and restricting users to perform optimally in an immersive environment. This paper reports on an experiment that aimed to test different speech-based designs for concurrent information communication. Two audio streams from two types of content were played concurrently to 34 users, in both a continuous or intermittent form, with the manipulation of a variety of spatial configurations (i.e. Diotic, Diotic-Monotic, and Dichotic). In total, 12 concurrent speech-based design configurations were tested with each user. The results showed that the concurrent speech-based information designs involving intermittent form and the spatial difference in information streams produce comprehensibility equal to the level achieved in sequential information communication.
Article
Full-text available
This chapter describes a hybrid prospective–retrospective theory of the way semantic context influences word recognition. This theory postulates the operation of two prospective priming mechanisms, expectancy and spreading activation, and one retrospective priming mechanism, a check for target-prime relatedness. There are four important ways in which hybrid prospective–retrospective theory of semantic priming promotes the understanding of the way semantic context influences the processes that mediate word recognition in lexical decision and pronunciation tasks. First, it uncovers the rather potent effects of a variable—the nonword ratio—that has heretofore been ignored in the analyses of semantic priming effects in the lexical decision task. Second, the theory suggests that an effect that has often been ignored and considered as a nuisance—the nonword facilitation effect—may actually be an important signature of the operation of a mechanism, the retrospective semantic matching mechanism, that can have a major influence on priming in the lexical decision task. Third, the hybrid prospective–retrospective theory of priming provides guidance as to whether a lexical decision or pronunciation task should be used to study the effects of semantic context. The fourth feature of the present theory is that it provides an integrative account of several diverse phenomena that have come to light in the heavily researched area of semantic context effects on word recognition.
Article
In semantic priming paradigms for lexical decisions, the probability that a word target is semantically related to its prime (the relatedness proportion) has been confounded with the probability that a target is a nonword, given that it is unrelated to its prime (the nonword ratio). This study unconfounded these two probabilities in a lexical decision task with category names as primes and with high- and low-dominance exemplars as targets. Semantic priming for high-dominance exemplars was modulated by the relatedness proportion and, to a lesser degree, by the nonword ratio. However, the nonword ratio exerted a stronger influence than did the relatedness proportion on semantic priming for low-dominance exemplars and on the nonword facilitation effect (i.e., the superiority in performance for nonword targets that follow a category name rather than a neutral XXX prime). These results suggest that semantic priming for lexical decisions is affected by both a prospective prime-generated expectancy, modulated by the relatedness proportion, and a retrospective target/prime semantic matching process, modulated by the nonword ratio.
Article
The present paper contrasts two theories about the use of semantic context information. On the one hand, the two-process theory maintains that subjects can use an automatic process alone or in conjunction with an attentional process to take advantage of a context. On the other hand, the strategy assumption maintains that subjects can select one of two semantic strategies in using a context. Two experiments examine the effect of different types of semantic context on third- and fifth-grade readers. The data, from a lexical-decision (word vs. nonword) task, strongly support the strategy assumption while disconfirming two-process theory./// [French] Cette étude met en contraste deux théories concernant l'utilisation d'information de contexte sémantique. D'une part, la théorie des deux processus soutient que les sujets peuvent utiliser un processus automatique et/ou un processus attentionnel pour profiter d'un contexte. D'autre part, l'hypothèse de stratégie soutient que les sujets peuvent sélectionner une des deux stratégies sémantiques en utilisant un contexte. Les deux essais examinent l'effet de différents types de contexte sémantique sur des lecteurs de neuvième et septième. Les données, à partir d'une tâche de décision lexicale (mots par rapport à des mots qui n'en sont pas), supportent avec force la supposition de stratégie tandis qu'elles déconfirment la théorie des deux procédés./// [Spanish] El presente artículo contrasta dos teorías sobre el uso de información de contexto semántico. Por un lado, la teoría de proceso doble declara que a individuos les es posible utilizar un proceso automático y/o un proceso de concentración para aprovechar del contexto. Por otro lado, la estrategia de suposición declara que a individuos les es posible seleccionar una o dos estrategias semánticas utilizando un contexto. Dos experimentos examinan los efectos de los diferentes tipos de contexto semántico en lectores de tercer y quinto grado. Los datos, derivados de una actividad de decisión lexical (palabra contra palabra sin significado) apoya firmemente la estrategia de suposición mientras desecha la teoría de proceso doble.