Acta Psychologica 172 (2017) 7176

Contents lists available at ScienceDirect

Acta Psychologica

journal homepage: www.elsevier.com/locate/actpsy

Sentence context inuences the subjective perception of foreign accents

Sara Incera a,, Amee P. Shah b, Conor T. McLennan a, Matthew T. Wetzel a
a
Language Research Laboratory, Department of Psychology, Cleveland State University, United States
b
School of Heath Sciences, Stockton University, United States

a r t i c l e i n f o a b s t r a c t

Article history: We examined whether sentence context (the predictability of the nal word) inuences listeners' ratings of for-
Received 12 May 2016 eign-accented words. Previous work has demonstrated that accent manipulations affect listeners' processing of
Received in revised form 10 November 2016 spoken language. We examined the converse of this relationship; whether context manipulations affect listeners'
Accepted 28 November 2016
perceptions of accents. If there is a bidirectional relationship, listeners should be more likely to rate an accent as
Available online 7 December 2016
strong when the accented word is not predicted by the sentence. In Experiment 1, the results revealed that par-
Keywords:
ticipants were signicantly more likely to rate words spoken by foreign-accented speakers as Strong Accent in
Foreign accent the unpredictable sentences when compared to the predictable sentences. Moreover, in Experiment 2, this effect
Misattribution was replicated and extended to a native speaker. These results support the idea that there is a bidirectional rela-
Sentence context tionship between language processing and perceptions of accents. We discuss the practical implications for for-
Predictability eign-accented speakers.
Speech perception 2016 Elsevier B.V. All rights reserved.

1. Introduction
According to the United Nations the number of international mi-
grants (those living outside their countries of birth) reached 244 million
in 2015, a 41% increase compared to 2000. Most of these immigrants
spoke a different language in their countries of origin, and thus likely
speak their new language with a foreign accent. Understanding how
these speakers communicate is an important topic with widespread
practical implications. Until very recently (Porretta, Kyrlinen &
Tucker, 2015), the vast majority of the literature in accentedness has fo-
cused on how accents inuence comprehension (Bent & Holt, 2013;
Bradlow & Bent, 2008; Bradlow & Bent, 2008; Clarke & Garrett, 2004;
Flege, Munro & MacKay, 1995; Levi, Winters & Pisoni, 2007; Shi &
Farooq, 2012; Ulbrich, 2013). In the present study we aim to understand
the converse of this relationship, how comprehension or lack thereof
inuences the subjective experience of perceiving a foreign accent. If
there is a bidirectional relationship between comprehension and the
perception of accents, foreign accents would make comprehension
more difcult, and at the same time, difculties in comprehension
would make the accents seem stronger.
We argue that people might misattribute communication difculties
(due to the sentence context) to the speaker's accent. If this is the case,
participants will subjectively rate the accent as stronger when the
accented stimulus is presented in a challenging contextual situation
(unpredictable sentence) than when it is presented in an easy contextu-
al situation (predictable sentence). When the stimulus is presented at
Corresponding author at: Language, Research Laboratory, Department of Psychology,
Cleveland State University, 2121 Euclid Avenue, Cleveland, OH 44115, United States.
E-mail address: saraincera@gmail.com (S. Incera).
the end of a sentence that it does not complete (e.g., The dog chased
our cat up the _ stamp) it will be rated as Strong Accent more often
than when the exact same foreign-accented recording is presented at
the end of a sentence it completes (e.g., He mailed the letter without a
_ stamp). The novel contribution of the present study is that we used
the exact same recordings in two contextual situations (predicted or
unpredicted nal word). If we nd differences in the ratings, we can at-
tribute these effects to the predictability of the sentence and not to char-
acteristics of the speaker. In this experiment, speaker-dependent factors
were not confounded with sentence context because all words (spoken
by all speakers) were counterbalanced across all sentences half the
time completing the sentence and half the time nishing a different sen-
tence in an unpredictable way (see the Appendix).
1.1. Accents inuence comprehension
The relationship between foreign-accented speech and comprehen-
sion has been a widely studied topic in spoken language research. As
stated above, much of the literature on this topic has focused on how
the perception of accents affects comprehension (Atagi & Bent, 2014,
2015; Bent & Holt, 2013; Bradlow & Bent, 2008; Bradlow & Bent,
2008; Clarke & Garrett, 2004; Flege et al., 1995; Levi et al., 2007; Shi &
Farooq, 2011; Ulbrich, 2013). Researchers have studied how regional
and foreign-accented speech inuences listeners' perception of spoken
words (Goslin, Duffy & Floccia, 2012; Hanulkov, Van Alphen, Van
Goch & Weber, 2012; Brunellire & Soto-Faraco, 2013; Jacewicz & Fox,
2015). Studies using map-drawing (Preston, 1986), card-sorting
(Tamasi, 2003), accent and dialect labeling, and categorizing (Shah,
2007; Clopper & Bradlow, 2008, 2009) tasks support the idea that lis-
teners attend to foreign accent and dialectal variations in speech.

http://dx.doi.org/10.1016/j.actpsy.2016.11.011
0001-6918/ 2016 Elsevier B.V. All rights reserved.
72 S. Incera et al. / Acta Psychologica 172 (2017) 7176
Moreover, listeners perceive non-native speakers as less credible (Lev-
Ari & Keysar, 2010), and listeners represent non-native speakers' lan-
guage with fewer details than that of native speakers (Lev-Ari &
Keysar, 2012). When listeners hear foreign-accented speech, processing
is slowed down, as indicated by an increase in processing time, a reduc-
tion in perceptual intelligibility and accuracy of the message, lowered
perception of overall signal comprehensibility, lowered phonetic and
word discrimination scores, a compromise in the ability to detect mis-
pronunciations, and a need for greater perceptual effort and cognitive
resources (e.g., Munro & Derwing, 1995a, 1995b; Weil, 2003; Shah,
Schmidt, Goral & Obler, 2005; Van Wijngaarden, 2001; Rogers, 1997;
Schmid & Yeni-Komshian, 1999). All of this evidence supports the idea
that foreign accents play a role in speech perception and processing.
1.2. Comprehension inuences accents
The literature has focused on how the perception of accents affects
how speech is understood. Nevertheless, the opposite direction of this
relationship, how comprehension inuences the subjective perception
of foreign accents, has not been studied in detail. It is important to high-
light that very few studies have examined how contextual information
inuences speech perception. Levi et al. (2007, p. 2328) argued that
speaker-independent factors [] can also affect the perceived degree
of foreign accent. Levi et al. (2007) evaluated the differences between
native and non-native speakers in two listening contexts (auditory-
only, auditory + orthography). These authors found that in the audito-
ry + orthography condition the perceived differences between native
and non-native speakers increased. Seeing the written word made par-
ticipants more aware of the differences between the native and non-na-
tive speakers. Levi et al. (2007) did not specically control for
production differences. Even though they argued that production differ-
ences could not account for all the observed interactions, a controlled
experiment will help to evaluate the isolated effect of speaker-indepen-
dent factors.
More recently, Porretta et al. (2015) examined how perception of
accentedness is affected by speaker-dependent and speaker-indepen-
dent variables simultaneously. In their study the speaker-dependent
factors were acoustic distances (how far the speaker is in relation to typ-
ical native-speaker values), while the speaker-independent variables
were lexical characteristics of the individual words (lexical frequency
for each word, number of phonological neighbors, and phonotactic
probability). These authors found that increased probability and activa-
tion of a particular item within the mental lexicon (e.g., higher lexical
frequency) resulted in lower perceived foreign accentedness. This nd-
ing is intriguing, participants associate native-like processing with eas-
iness, and they rate the accent as less strong when it is easier to
process even when the easiness comes from characteristics of the
words that have nothing to do with the accent. Porretta et al. (2015)
concluded that both acoustic (speaker-dependent) and lexical (speak-
er-independent) properties are involved in the matching of a particular
token to the representation of what constitutes a native-like production.
1.3. Sentence context
Semantic predictability facilitates word recognition during language
processing. Researchers report that high-predictability sentences dis-
play lower error rates and faster response times than low-predictability
sentences (Pisoni, Manous & Dedina, 1987). Luke and Christianson
(2012) argued that during online language processing highly specic
predictions preactivate some features of upcoming words. Importantly,
semantic predictability has been shown to interact with familiar dia-
lects. Clopper (2012) observed larger semantic predictability benets
in the perception of words produced by speakers with more familiar di-
alects than speakers with less familiar dialects. In production, vowel du-
ration was reduced in sentences with high-semantic predictability for
Southern and Northern talkers of American English (Clopper &
Pierrehumbert, 2008). If the relationship between semantic predictabil-
ity and foreign accents is bidirectional, the predictability of the nal
word will inuence judgments about the speaker. There have been a
few studies (Behrman & Akhund, 2013; Derwing & Munro, 1997;
Kennedy & Tromovich, 2008) that have touched on whether sentence
context affects accents ratings independently of the acoustic
information.
Derwing and Munro (1997) attempted to correlate intelligibility,
comprehensibility, and accentedness by asking participants to listen to
a short story told by speakers with foreign accents. Participants can dis-
tinguish between speaker-dependent (accentedness) and context-de-
pendent (comprehensibility) factors, but these factors inuence one
another. Their ndings suggest that accentedness, comprehensibility,
and intelligibility scores are clearly related but are not equivalent
(Derwing & Munro, 1997). Behrman and Akhund (2013) found a signif-
icant main effect of semantic context on accentedness, comprehensibil-
ity, and intelligibility. Accents were perceived to be stronger, and both
comprehensibility and intelligibility were worse, in semantically anom-
alous contexts. Finally, Kennedy and Tromovich (2008) studied
whether semantic context affects perception of foreign accents. These
researchers found that sentences that were more difcult to understand
(semantically meaningless) were rated as more accented for the non-
native speakers.
Studies to date do not rule out the possibility that accents may in-
deed become stronger as speakers nd complex material more difcult
to pronounce. Our experiment adds to the literature by explicitly
distinguishing between speaker-dependent and speaker-independent
factors (a confound in these previous studies). We used the same re-
cording in two different sentence contexts in order to ensure that
speaker-dependent factors are not inuencing the results. The goal is
to study the unique inuence of speaker-independent factors (i.e., sen-
tence predictability).
1.4. Mixed models
In the present study, participants rated the level of accentedness of
spoken words by clicking on the Weak Accent or on the Strong Ac-
cent response option with a computer mouse (see Fig. 1). We
implemented the mouse-tracking paradigm (Spivey, Grosjean &
Knoblich, 2005) using the computer program MouseTracker (Freeman
& Ambady, 2010). We used mixed logit models (Jaeger, 2008) to evalu-
ate participants' ratings of the foreign-accented words (0 Weak
Accent, 1 Strong Accent). Their responses are dichotomous, therefore
mixed logit models are the best approach to evaluate the effects of Ac-
cent and Sentence Context. Moreover, using mixed models it is possible
to include participants and words crossed at the same level of sampling.
Fig. 1. This is the screen that participants viewed during the experiment. They had to click
START to hear each sentence, and then move the mouse toward one of the response
alternatives (Weak Accent or Strong Accent).
 S. Incera et al. / Acta Psychologica 172 (2017) 7176
Following Linck and Cunnings (2015, p.199), we specied the depen-
dent variable as a binary outcome with the family = binomial argu-
ment, and we included the variables Accent and Context as random
slopes by participants. We included the same xed and random effects
when evaluating participants' ratings across both experiments.1 All
analyses were performed using lme4 version 1.112 in R version 3.3.1.
1.5. Predictions
Our experiments extend previous research by studying the inuence
of speaker-independent factors in isolation. In the present experiment
we used the exact same audio recording of a word in two different sen-
tence contexts.2 By using the same recording, we were able to control
for all speaker-dependent factors. We argue that the impact of speak-
er-independent factors, in this case sentence context, will affect lis-
teners' perception of a foreign accent. In particular, we are interested
in studying how the predictability of the accented word (predictable
or unpredictable sentence context) inuences the subjective experience
of rating an accent as Strong or Weak. We argue that when participants
cannot predict the word, accents are more likely to be rated as being
strong, relative to when those exact same audio les are heard in a pre-
dictable context.
We predict that listeners will rate the nal word of a sentence as
having a strong accent less often when the word completes a sentence
(predictable) relative to when the same recording of that word does
not complete the sentence (unpredictable). If there is a bidirectional re-
lationship between accents and comprehension, speaker-independent
factors that affect comprehension should play a role in accent percep-
tion. We will observe more Strong Accent ratings for the unpredicted
nal word than for the predicted nal word, despite using the exact
same recordings. This nding would suggest that listeners are attribut-
ing (or more accurately, misattributing) the uency with which they
process the spoken word (which is due to the sentence predictability)
to the strength of the speaker's accent. Moreover, we predict that this
misattribution will extend to native speakers (Experiment 2). If the ef-
fect is replicated in a native speaker, it would further suggest that sen-
tence context inuences perception above and beyond speaker
characteristics. Finding a sentence context effect in native speakers
would strongly support the argument that it is necessary to include
speaker-independent factors in models of speech perception.
We should note that a longstanding debate in research on spoken
word recognition and speech perception is whether higher levels can
inuence listeners' perception at lower levels (see e.g., McClelland,
Mirman & Holt, 2006). Although participants in the current study are
providing accent ratings, as opposed to a word recognition or speech
perception task, if the above predictions are borne out, the data may
be more consistent with an interactionist approach to this debate in
which higher levels (in this case semantic context) can inuence
lower levels (in this case ratings about the speech signal). Alternatively,
should the predicted results fail to emerge, then such results may be
more consistent with autonomous accounts in which lower levels are
uninuenced by higher levels.
1
m.Full glmer(Rating ~ Accent Context + (Accent Context|Subject) + (1|Word),
data = Experiment, family = binomial, control = record participants' responses (see also Incera & McLennan, 2016a;
glmerControl(optimizer = bobyqa))
2 c
The present investigation was building off previous work. Shah and M Lennan did a
button-press sentence context study (2008). In this previous study there was no effect
in accent rating responses; instead the effect was obtained in reaction time (RT) differ-
ences (i.e., longer RTs to rate accents for words in an unpredictable sentence context).
73
2. Experiment 1: Six foreign speakers
2.1. Method
2.1.1. Participants
Twenty-four participants from the Cleveland State University com-
munity participated in the experiment. This sample size was deter-
mined in order to completely counterbalance six speakers and two
types of sentences (12 versions of the experiment), while
counterbalancing the position of the two response options on the
screen. Participants received research participation credit for a course
requirement. Participants were right-handed native speakers of Ameri-
can English with no reported history of speech or hearing disorders.
2.1.2. Materials
Stimuli consisted of 24 sentences (see Appendix) selected from the
sentence corpus by Bloom and Fischler (1980) with the following
criteria: The sentences contained a highly-predictable nal word (e.g.,
They sat together without speaking a singleword), used active
voice, and were balanced for the overall sentence-length (69 words)
and the number of syllables in the nal word (1 or 2). Additionally, six
practice sentences were recorded. The unpredictable sentences were
created by combining each carrier sentence with the nal word of one
of the other sentences (in the Appendix the word in parenthesis). The
combination of each carrier portion with a nal word resulted in one
predictable sentence (when paired with its original word) and one un-
predictable sentences (when paired with the last word of one of the
other sentences).
The carrier portion of the sentence was recorded separately and by a
different speaker than the target words. All target words were recorded
in isolation by different speakers; the speakers never had access to the
sentences in order to control for possible coarticulation. The carrier por-
tion of the sentence was produced in a different voice (native speaker)
in order to ensure that the raters were only attending to the
accentedness in the nal word (foreign speakers). In addition, there
was a gender difference between the native voice (female speaker)
and the foreign-accented voices (male speakers) in order to make the
difference clear. The incomplete sentences (without the nal word)
were recorded by a speaker born and raised in the Northeast Ohio re-
gion who had a standard Midwestern accent (a native accent to our par-
ticipants). The nal words were recorded by six male foreign speakers:
two Hindi, two Chinese, and two Arabic. The mean durations of the 24
nal words produced by the foreign speakers were not signicantly dif-
ferent across speakers, and the mean duration across words was
544 ms. All words were recorded in a quiet room at Cleveland State Uni-
versity, and then normalized and equated on mean amplitude.
2.1.3. Design
There were two within participants' factors: accent (Hindi, Chinese,
Arabic) and sentence context (predictable or unpredictable). Each par-
ticipant responded to 24 sentences (12 predictable, 12 unpredictable).
The words were completely counterbalanced between the six speakers
and the two types of sentences, resulting in 12 versions of the experi-
ment. Across participants every speaker pronounced every word in a
predictable and unpredictable sentence context. For each trial the carri-
er sentence was presented rst, followed by the target word (complet-
ing or not that particular sentence).
2.1.4. Procedure
We used MouseTracker (for a detailed explanation of the mouse
tracking method see Freeman & Ambady, 2010; Spivey et al., 2005) to
Incera & McLennan, 2016b; Yamamoto, Incera, & McLennan, 2016;
Krestar, Incera, & McLennan, 2013). Participants were instructed to
click on Weak Accent or Strong Accent as quickly and accurately as
possible (see Fig. 1). Listeners were not given any instructions regarding
74 S. Incera et al. / Acta Psychologica 172 (2017) 7176
what weak and strong accent meant, and they were not told that some
words would be predictable and other words would be unpredictable.
We did not include a no accent option because the purpose of the ex-
periment was to observe changes within the level of accentedness.
The position of the two response options was counterbalanced across
participants.
There were four practice trials followed by 24 experimental trials. A
given trial proceeded as follows: Participants clicked START and were
immediately presented with a sentence, spoken by a native female talk-
er, without the nal word. After a 100 ms delay (this small pause was
the cue that informed participants they should get ready to move the
mouse) the nal word of the sentence was presented spoken by one
of the six foreign speakers. This nal word either completed the sen-
tence (predicted word) or caused the sentence to be meaningless
(unpredicted word). Sentences and nal words were presented binau-
rally over headphones. Participants were instructed to listen to the
sentences carefully, and to make their accent ratings based on only
the nal word of each sentence (spoken by the male foreign talkers)
and not based on the rest of the sentence leading up to the nal word
(spoken by the female native talker). After the participant responded,
the next trial was initiated.
2.2. Results
There were a total of 24 trials per participant (n = 24), for a grand
total of 576 responses across participants. Participants' ratings3 were di-
chotomous (Strong-Weak), thus we analyzed the data using logit mixed
models (Jaeger, 2008). We evaluated the xed effects of Accent (Hindi,
Chinese, Arabic), Sentence Context (predictable, unpredictable), and
the Accent by Context interaction using model comparisons. We includ-
ed the random effects of participants and items crossed at the same
level of sampling. In addition, we included the variables Accent and
Context as random slopes by participants. Model comparisons indicated
that the main effect of Accent and the Accent by Context interaction
were not signicant, but the main effect of Context signicantly im-
proved model t (2 (1) = 9.06, p = 0.003). Consistent with our predic-
tion, words in the unpredictable sentences were rated as having a
Strong Accent more often (Estimate = 3.80, SE = 0.72, z = 5.25,
p b 0.001) than words in the predictable sentences (see Fig. 2).
3. Experiment 2: A native and a foreign speaker
In Experiment 1 we found evidence that participants rate words spo-
ken by foreign-accented speakers as more accented when the words are
not predicted by their sentences (sentence context inuenced partici-
pants' ratings of foreign-accented words). The motivation of Experi-
ment 2 is to evaluate whether this effect extends to native speech.
3.1. Method
3.1.1. Participants
Twenty-four new participants were recruited from the Cleveland
State University community. Participants received partial credit for a
course requirement. Participants were right-handed native speakers of
American English with no reported history of speech or hearing
disorders.
3.1.2. Materials
Stimuli consisted of the same 24 sentences (see the Appendix). The
incomplete sentences (without the nal word) were the same record-
ings from Experiment 1. The nal words were recorded again by two
new male speakers: a native (Midwestern) and a foreign-accented
3
In the present study, mouse-tracking data were also collected. The results did not add
to the analyses reported in the paper, so we chose not to report them here. However, the
analyses are accessible to the interested reader as supplementary materials.
Fig. 2. The number of responses rated Strong Accent for the predictable and
unpredictable sentence context conditions across all three accents.
(Hindi) speaker. The mean durations of the 24 nal words were
401 ms for the words produced by the native and 502 ms for the
words produced by the foreign-accented speaker. All words were re-
corded in a quiet room at Cleveland State University, and then normal-
ized and equated on mean amplitude.
3.1.3. Design
In line with the analysis of Experiment 1, we evaluated two within-
participants' factors: accent (native or foreign) and sentence context
(predictable or unpredictable). In this version of the experiment, six
sentences (out of 12 per condition) were completed by the nal word
spoken by the native speaker, and the other six were completed by
the nal word spoken by the foreign speaker. The words were
completely counterbalanced between speakers and types of sentences,
so across participants every speaker pronounced every word in both
sentence contexts.
3.1.4. Procedure
The procedure was identical to the one used in Experiment 1.
3.2. Results
There were a total of 24 trials per participant, for a grand total of 576
responses across participants. Participants' responses were dichoto-
mous (Strong-Weak), thus we analyzed the data using logit mixed
models (Jaeger, 2008). We evaluated the xed effects of Accent (foreign,
native), Sentence Context (predictable, unpredictable), and the Accent
by Context interaction. We included the random effects of participants
and items crossed at the same level of sampling. In addition, we includ-
ed the variables Accent and Context as random slopes by participants.
Model comparisons indicated that the main effect of Accent (2 (1) =
27.59, p b 0.001), the main effect of Context (2 (1) = 7.41, p =
0.006), and the Accent by Context interaction (2 (1) = 3.78, p =
0.051) improved model t. Not surprisingly, the foreign speaker was
rated as having a stronger accent than the native speaker (Estimate =
9.03, SE = 1.35, z = 6.68, p b 0.001). As in Experiment 1, participants
rated the unpredictable sentences as having a stronger accent than the
predictable sentences (Estimate = 2.40, SE = 0.74, z = 3.22, p =
0.001). Interestingly, the interaction emerged because the effect of con-
text unpredictable sentences being rated as having a Strong Accent
more often, see Fig. 3 was more pronounced in the native than the for-
eign-accented speaker (Estimate = 11.27, SE = 2.01, z = 5.58,
p b 0.001). In sum, these results indicate that participants rated a spoken
word as having a Strong Accent more often in the context of an unpre-
dictable sentence than in the context of a predictable sentence. Impor-
tantly, this effect extended to the native speaker.
 S. Incera et al. / Acta Psychologica 172 (2017) 7176
Fig. 3. The number of responses rated Strong Accent in both the predictable and
unpredictable sentence context conditions, for the native and foreign-accented speakers.
4. Discussion
To date, the majority of studies investigating accentedness have fo-
cused on manipulating various speaker-dependent factors (e.g., talker dif-
ferences) to determine whether these manipulations affect listeners'
comprehension. In this investigation, we aimed to understand the con-
verse of this relationship; whether manipulating speaker-independent
factors (that help or hurt comprehension) inuences participants' ratings
of foreign-accented speech. Specically, listeners were asked to judge the
degree of accentedness of the nal word in either predictable or unpre-
dictable sentence contexts. To our knowledge, this is the rst study to ma-
nipulate sentence context and observe signicant effects in accent ratings
while using the exact same recording. In this study we obtained evidence
that linguistic context inuences accent ratings. In Experiment 1, the re-
sults revealed that participants were more likely to rate a spoken word
as having a Strong Accent in the context of an unpredictable sentence
than when the same recording was presented in a predictable sentence
context. Importantly, in Experiment 2, this effect was replicated and ex-
tended to a native speaker. Listeners used the context as part of their
basis for making accent ratings, and were more likely to rate an accent
as strong when presented in an unpredictable sentence.
Researchers have argued that speaker-independent factors can in-
teract with speaker-dependent factors (Levi et al., 2007; Porretta et al.
2015). Our study went one step further, measuring the inuence of
speaker-independent factors in isolation, while controlling for speak-
er-dependent factors. It is remarkable that the exact same recording
(same word, same speaker) was judged as a stronger accent when pre-
sented in an unpredictable sentence context than when presented in a
predictable sentence context. When making a judgment of foreign ac-
cents, listeners do not simply access knowledge of the native language
phonology; listeners also use contextual information to guide their sub-
jective impressions. Importantly, the effect of sentence context was also
found for the native speaker. This nding supports the notion that this
effect is independent of the actual characteristics of the speaker. The
fact that the effect of sentence context inuenced the ratings of native
speakers raises the question of what is really being measured,
accentedness or processing effort. Our results suggest that processing
effort is being captured by the judgment of accentedness. The unpre-
dictable context made overall comprehension more effortful, which in
turn inuenced participants' ratings. It remains to be determined to
what extent accentedness and processing effort are separable. Our re-
sults follow those of Derwing and Munro (1997), who indicate that
accentedness and comprehensibility are related constructs, and support
a bidirectional relationship between comprehension and accent ratings.
We return to a point raised in the Introduction regarding the debate
between interactionist and autonomous models of spoken word
75
recognition and speech perception. This debate is often centered around
the question of whether the lexical level can inuence perception of
stimuli at the pre- (or sub-) lexical level (e.g., Norris, McQueen &
Cutler, 2016). Distinctions have been made between online perception
and learning (Bowers & Davis, 2004). The question remains a hot topic
(Norris et al., 2000) and source of great debate (Gow & Olson, 2016). Al-
though participants in the current study were asked to make accent rat-
ings, as opposed to a spoken word recognition or speech perception
task, the current results may be more consistent with interactionist
models. We add to the body of literature investigating top-down effects
on lower levels in which the higher level involves semantic and prag-
matic (Rohde & Ettlinger, 2012), meaning (Tyler, Voice & Moss, 2000),
and syntactic sentential context (Fox & Blumstein, 2016).
Our ndings have practical implications because they show that par-
ticipants can misattribute their own communication difculties to the
speaker characteristics. Listeners are incorrectly blaming the accent
instead of realizing that there are contextual difculties. We are not, of
course, arguing that in everyday situations participants would perceive
native speakers as foreigners. What we are saying is that listeners can
misattribute the source of some communications difculties in this
case the lack of an appropriate sentence context can be misattributed
as a strong foreign accent and that this effect occurs even when listen-
ing to native-accented speech. Future research is necessary in order to
better understand if, outside the laboratory, listeners' interpretations
are different with different speakers. It is likely that in the presence of
a native speaker, the listener would be more likely to look for alternative
sources of communication challenges (e.g., difcult information, back-
ground noise, etc.) on which to attribute any perceptual difculties. In
the presence of a foreign speaker it might be easier to misattribute or
blame the difculties on the accent. The present ndings highlight
that these effects are not modulated by the actual signal, but instead
are modulated by contextual inuences (the predictability of sentence
context). When contextual inuences make communication challeng-
ing, a foreign accent might be an easy target for blame. These effects
might be particularly salient in higher education, where the information
presented is typically difcult, or in medical settings, where jargon can
be complicated to understand.
We argue that in everyday situations, people might underestimate
contextual information and, as a consequence of doing so, misattribute
communication challenges to foreign speakers. Importantly, foreign-ac-
cent perceptions are not only inuenced by the speaker or the listener
characteristics, but also by the surrounding context in which the commu-
nication takes place. These results have implications for models and the-
ories of spoken word recognition. In particular, it may no longer be
sufcient to suggest that accentedness affects the way in which listeners
process speech. In addition, models and theories of spoken word recogni-
tion will need to account for the circumstances under which speaker-in-
dependent factors (e.g., sentence context) affects listeners' subjective
perceptions of foreign accents. Future research should further consider
the implications of the bidirectional relationship between comprehension
and accent perception, and study the specic contextual circumstances
that can shape the way in which foreign accented speakers are perceived
(see Pisoni & McLennan, 2015). Models of spoken language need to in-
clude speaker-independent factors, in addition to speaker-dependent fac-
tors, when accounting for communication challenges.
In conclusion, we observed that when a word was presented in an un-
predictable context participants rated the accent as stronger. Importantly,
this effect was also observed in a native speaker. These results provide
support for the idea that speaker-independent variables (e.g., sentence
predictability) inuence participants' ratings of the strength of an accent.
Appendix A
He mailed the letter without a _ stamp (tree)
He scraped the cold food from his _ plate (dinner)
The dough was put in the hot _ oven (knife)
76 S. Incera et al. / Acta Psychologica 172 (2017) 7176
She called her husband at his _ ofce (hair)
The children went outside to _ play (tip)
Her new shoes were the wrong _ size (word)
The dog chased our cat up the _ tree (stamp)
Jeff was sent to bed without _ dinner (plate)
Father carved the turkey with a _ knife (oven)
She went to the salon to color her _ hair (ofce)
The rude waiter was not given a _ tip (play)
They sat together without speaking a single _ word (size)
Joan fed her baby some warm _ milk (window)
He loosened the tie around his _ neck (letter)
John swept the oor with a _ broom (mother)
The paint turned out to be the wrong _ color (cotton)
The child was born with a rare _ disease (money)
He hung her coat in the _ closet (dishes)
Tim threw a rock and broke the _ window (milk)
Hal wrote her a love _ letter (neck)
She was named after her _ mother (broom)
David's shirt was made of _ cotton (color)
Starting a business takes a lot of _ money (disease)
After dinner they washed the _ dishes (closet)
Appendix B. Supplementary data
Supplementary data to this article can be found online at http://dx.
doi.org/10.1016/j.actpsy.2016.11.011.
References
Atagi, E., & Bent, T. (2014). Auditory free classication of native and nonnative speech by
nonnative listeners. Applied PsychoLinguistics, 123.
Atagi, E., & Bent, T. (2015). Relationship between listeners' nonnative speech recognition
and categorization abilities. The Journal of the Acoustical Society of America, 137,
4450.
Behrman, A., & Akhund, A. (2013). The inuence of semantic context on the perception of
Spanish-accented American English. Journal of Speech, Language, and Hearing
Research, 56, 15671578.
Bent, T., & Holt, R. F. (2013). The inuence of talker and foreign-accent variability on spo-
ken word identication. The Journal of the Acoustical Society of America, 133,
16771686.
Bloom, P. A., & Fischler, I. (1980). Completion norms for 329 sentence contexts. Memory
and Cognition, 8, 631642.
Bowers, J. S., & Davis, C. J. (2004). Is speech perception modular or interactive? Trends in
Cognitive Sciences, 8, 35.
Bradlow, A. R., & Bent, T. (2008). Perceptual adaptation to non-native speech. Cognition,
106, 707729.
Brunellire, A., & Soto-Faraco, S. (2013). The speakers' accent shapes the listeners' phono-
logical predictions during speech perception. Brain and Language, 125, 8293.
Clarke, C. M., & Garrett, M. F. (2004). Rapid adaptation to foreign-accented English. The
Journal of the Acoustical Society of America, 116, 36473658.
Clopper, C. G. (2012). Effects of dialect variation on the semantic predictability benet.
Language & Cognitive Processes, 27, 10021020.
Clopper, C. G., & Bradlow, A. R. (2008). Perception of dialect variation in noise: Intelligibil-
ity and classication. Language and Speech, 51, 175198.
Clopper, C. G., & Bradlow, A. R. (2009). Free classication of American English dialects by
native and non-native listeners. Journal of Phonetics, 37, 436451.
Clopper, C. G., & Pierrehumbert, J. B. (2008). Effects of semantic predictability and regional
dialect on vowel space reduction. The Journal of the Acoustical Society of America, 124,
16821688.
Derwing, T. M., & Munro, M. J. (1997). Accent, intelligibility and comprehensibility: Evi-
dence from four L1 s. Studies in Second Language Acquisition, 19, 116.
Flege, J. E., Munro, M. J., & MacKay, I. R. (1995). Factors affecting strength of perceived for-
eign accent in a second language. Journal of the Acoustical Society of America, 97,
31253134.
Fox, N. P., & Blumstein, S. E. (2016). Top-down effects of syntactic sentential context on
phonetic processing. Journal of Experimental Psychology: Human Perception and
Performance, 42, 730741.
Freeman, J. B., & Ambady, N. (2010). MouseTracker: Software for studying real-time men-
tal processing using a computer mouse-tracking method. Behavior Research Methods,
42, 226241.
Goslin, J., Duffy, H., & Floccia, C. (2012). An ERP investigation of regional and foreign ac-
cent processing. Brain and Language, 122, 92102.
Gow, D. W., & Olson, B. B. (2016). Using effective connectivity analyses to understand pro-
cessing architecture: Response to commentaries by Samuel, Spivey and McQueen,
Eisner, and Norris. Language, Cognition, and Neuroscience, 31, 869875.
Hanulkov, A., Van Alphen, P. M., Van Goch, M. M., & Weber, A. (2012). When one
person's mistake is another's standard usage: The effect of foreign accent on syntactic
processing. Journal of Cognitive Neuroscience, 24, 878887.
Incera, S., & McLennan, C. T. (2016a). Mouse tracking reveals that bilinguals behave like
experts. Bilingualism: Language and Cognition, 19, 610620.
Incera, S., & McLennan, C. T. (2016b). The time course of within and between language in-
terference in bilinguals. International Journal of Bilingualism, 112.
Jacewicz, E., & Fox, R. A. (2015). The effects of dialect variation on speech intelligibility in a
multitalker background. Applied PsychoLinguistics, 36, 729746.
Jaeger, T. F. (2008). Categorical data analysis: Away from ANOVAs (transformation or not)
and towards logit mixed models. Journal of Memory and Language, 59, 434446.
Kennedy, S., & Tromovich, P. (2008). Intelligibility, comprehensibility, and accentedness
of L2 speech: The role of listener experience and semantic context. The Canadian
Modern Language Review, 64, 459489.
Krestar, M., Incera, S., & McLennan, C. T. (2013). Using mouse tracking to examine the
time course of an auditory lexical decision task. Ohio Psychologist, 60, 2932.
Lev-Ari, S., & Keysar, B. (2010). Why don't we believe non-native speakers? The inuence
of accent on credibility. Journal of Experimental Social Psychology, 46, 10931096.
Lev-Ari, S., & Keysar, B. (2012). Less-detailed representation of non-native language: Why
non-native speakers' stories seem more vague. Discourse Processes, 49, 523538.
Levi, S. V., Winters, S. J., & Pisoni, D. B. (2007). Speaker-independent factors affecting the
perception of foreign accent in a second language. Journal of the Acoustical Society of
America, 23272338.
Linck, J. A., & Cunnings, I. (2015). The utility and application of mixed-effects models in
second language research. Language Learning, 65, 185207.
Luke, S. G., & Christianson, K. (2012). Semantic predictability eliminates the transposed-
letter effect. Memory & Cognition, 40, 628641.
McClelland, J. L., Mirman, D., & Holt, L. L. (2006). Are there interactive processes in speech
perception? Trends in Cognitive Sciences, 10, 363369.
Munro, M. J., & Derwing, T. M. (1995a). Foreign accent, comprehensibility, and intelligibil-
ity in the speech of second language learners. Language Learning, 45, 7397.
Munro, M. J., & Derwing, T. M. (1995b). Processing time, accent, and comprehensibility in
perception of native and foreign-accented speech. Language and Speech, 38, 289306.
Norris, D., McQueen, J. M., & Cutler, A. (2000). Merging information in speech recognition:
Feedback is never necessary. Behavioral and Brain Sciences, 23, 299370.
Norris, D., McQueen, J. M., & Cutler, A. (2016). Predication, Bayesian inference and feed-
back in speech recognition. Language, Cognition, and Neuroscience, 31, 414.
Pisoni, D. B., & McLennan, C. T. (2015). Spoken word recognition: Historical roots, current
theoretical issues, and some new directions. In G. Hickok, & S. L. Small (Eds.), Neuro-
biology of language (pp. 239253). San Diego: Elsevier.
Pisoni, D. B., Manous, L. M., & Dedina, M. J. (1987). Comprehension of natural and synthet-
ic speech: Effects of predictability on the verication of sentences controlled for intel-
ligibility. Computer Speech & Language, 2, 303320.
Porretta, V., Kyrlinen, A. J., & Tucker, B. V. (2015). Perceived foreign accentedness:
Acoustic distances and lexical properties. Attention, Perception, & Psychophysics, 114.
Preston, D. R. (1986). Five visions of America. Language in Society, 15, 221240.
Rogers, C. L. (1997). Intelligibility of Chinese-accented English (Doctoral dissertation). Illi-
nois: Indiana University.
Rohde, H., & Ettlinger, M. (2012). Integration of pragmatic and phonetic cues in spoken
word recognition. Journal of Experimental Psychology: Learning, Memory, and
Cognition, 38, 967983.
Schmid, P. M., & Yeni-Komshian, G. H. (1999). The effects of speaker accent and target
predictability on perception of mispronunciations. Journal of Speech, Language, and
Hearing Research, 42, 5664.
Shah, A. P. (2007). Accuracy and response times in identifying foreign accents: Are lis-
teners all over the map? Journal of the Acoustical Society of America, 122, 243.
Shah, A. P., & McLennan, C. T. (2008). The role of foreign-accentedness in lexical process-
ing. Proceedings of the Meeting of the Acoustical Society of America, 2, 113.
Shah, A. P., Schmidt, B. T., Goral, M., & Obler, L. K. (2005). Age effects in processing bilin-
guals' accented speech. In J. Cohen, K. McAlister, K. Rolstad, & J. MacSwan (Eds.), ISB4:
Proceedings of the 4th International Symposium on Bilingualism (pp. 21152121). Som-
erville, MA: Cascadilla Press.
Shi, L. F., & Farooq, N. (2012). Bilingual listeners' perception of temporally manipulated
English passages. Journal of Speech, Language, and Hearing Research, 55, 125138.
Spivey, M. J., Grosjean, M., & Knoblich, G. (2005). Continuous attraction toward phonolog-
ical competitors. Proceedings of the National Academy of Sciences, 102, 1039310398.
Tamasi, S. L. (2003). Cognitive patterns of linguistic perceptions (Doctoral dissertation). Ath-
ens: University of Georgia.
Tyler, L. K., Voice, K., & Moss, H. E. (2000). The interaction of meaning and sound in spo-
ken word recognition. Psychonomic Bulletin & Review, 7, 320326.
Ulbrich, C. (2013). German pitches in English: Production and perception of crossvarietal
differences in L2. Bilingualism: Language and Cognition, 16, 397419.
Van Wijngaarden, S. J. (2001). Intelligibility of native and non-native Dutch speech.
Speech Communication, 35, 103113.
Weil, S. (2003). The impact of perceptual dissimilarity on the perception of foreign-accented
speech (Doctoral dissertation). Ohio: Ohio State University.
Yamamoto, N., Incera, S., & McLennan, C. T. (2016). A reverse Stroop task with mouse
tracking. Frontiers in Psychology, 7, 670.