Conference PaperPDF Available

LANGUAGE ATTITUDES AFFECT PERCEIVED INTELLIGIBILITY, PROFICIENCY, AND ACCENTEDNESS OF NON-NATIVE SPEECH

Authors:

Abstract and Figures

Previous research [7], showed that Mandarin speakers of English modified acoustic properties of their English speech as a factor of both the interlocutor (native vs. non-native speakers of English) and their own attitudes towards Mandarin and English. The present study investigates whether these acoustic modifications are perceptible to native speakers of English. Seventy-two native English listeners rated short English speech samples from twenty-four Mandarin learners with respect to speaker's intelligibility, proficiency, and accentedness, on a 7-point scale. The results showed that the interlocutor condition was not reflected in listeners' ratings. However, speakers' attitudes significantly predicted listeners' ratings. Participants who were more positively oriented towards Mandarin than English were perceived as less intelligible, less proficient, and more accented. The results suggest that the effects of language attitudes on second language speech are salient and perceptible to native listeners.
Content may be subject to copyright.
LANGUAGE ATTITUDES AFFECT PERCEIVED INTELLIGIBILITY,
PROFICIENCY, AND ACCENTEDNESS OF NON-NATIVE SPEECH
Amy E. Hutchinson1, Joshua D. Weirick1, Suzy Ahn2, Olga Dmitrieva1
1Purdue University, 2University of California, Los Angeles
hutchi25@purdue.edu, jweiric@purdue.edu, suzyahn@ucla.edu, odmitrie@purdue.edu
ABSTRACT
Previous research [7], showed that Mandarin speakers
of English modified acoustic properties of their
English speech as a factor of both the interlocutor
(native vs. non-native speakers of English) and their
own attitudes towards Mandarin and English. The
present study investigates whether these acoustic
modifications are perceptible to native speakers of
English. Seventy-two native English listeners rated
short English speech samples from twenty-four
Mandarin learners with respect to speaker’s
intelligibility, proficiency, and accentedness, on a 7-
point scale. The results showed that the interlocutor
condition was not reflected in listeners’ ratings.
However, speakers’ attitudes significantly predicted
listeners’ ratings. Participants who were more
positively oriented towards Mandarin than English
were perceived as less intelligible, less proficient, and
more accented. The results suggest that the effects of
language attitudes on second language speech are
salient and perceptible to native listeners.
Keywords: Non-native speech, language attitudes,
intelligibility, proficiency, accentedness
1. INTRODUCTION
The acquisition of a second language (L2) requires
much more than learning skills because of its inherent
social nature. In order to be successful in the
acquisition of an L2, the learner must be willing to
adapt and alter their self-image, and previous research
suggests that learners who approach L2 acquisition
with a more positive attitude are likely to achieve
higher ultimate attainment in the language ([6, 10,
21], inter alia).
Among other aspects of linguistic competence,
acquisition of L2 pronunciation can be affected by
learners’ attitudes [8, 12, 13, 14, 15, 20].
Pronunciation is subject to a number of factors such
as long-term transfer effects, fossilization, and age-
effects, which therefore makes it challenging for the
learner to achieve target-like proficiency [16].
Nevertheless, research has shown that a positive
attitude towards the L2 aids in achieving near target-
like pronunciation [8, 12, 15, 20].
One study that demonstrated the effect of speaker
attitude on non-native speech [7] examined the
speech of Mandarin speakers of English conversing
with interlocuters from different L1 backgrounds
(Mandarin, Russian, and English). When addressing
native English speakers, Mandarin speakers who
reported being more English-oriented (had a positive
attitude towards English), utilized a more
hyperarticulated vowel space, faster articulation rate,
and higher pitch compared to speakers who were
more Mandarin-oriented.
While [7] found that English-oriented Mandarin
speakers made modifications to their English speech,
it is unclear whether those adaptations were
implemented with a purpose in mind. One possibility
is that participants aimed to make their speech more
intelligible or less accented when addressing English-
speaking interlocutors. In this case, multiple acoustic
properties of their speech would reflect this goal, in
addition to those explored in [7] (vowel space
expansion, global pitch and articulation rate).
Moreover, if the acoustic modifications are salient
enough, human listeners would be able to detect them
in a perceptual judgement experiment.
The current study investigates whether the
interlocutor factor (L1 background of the person the
speaker is addressing) and speakers’ language
attitudes have an effect on native listener judgments
of speakers’ intelligibility, accentedness, and
proficiency in English. Evidence of such effects
would suggest that non-native speakers modify the
acoustic properties of their speech with specific
listeners and specific goals in mind, and that their
ability or wiliness to implement such modifications
depends on their language attitudes.
2. METHODS
2.1. Participants
Seventy speakers of American English (mean age
20.7 years, 7 male and 63 female) participated in this
study as raters. Raters were recruited among students
enrolled at a major Midwestern American University
and were compensated for their time with course
credit.
2233
2.2. Materials
The experimental stimuli were short sound clips
extracted from longer recordings of 24 Mandarin
speakers completing a map task [1]. In each
recording, a speaker gave directions to a confederate
using a map with labeled landmarks. All speakers
were native speakers of Mandarin from the same
dialectal area (Beijing and Northern China). Three
different confederates participated in the task: a
native speaker of Mandarin, a native speaker of
Midwestern American English, and a native speaker
of Russian. Twenty-four Mandarin speakers
completed the task three times, once with each
confederate. After completing the direction-giving
task, each of the speakers took a language background
questionnaire, which featured questions regarding
their age of arrival (AOA), length of residence in the
US (LOR), onset of L2 acquisition, and years of
English schooling. The questionnaire also included a
section on language attitudes adopted from the
Bilingual Language Profile questionnaire [3].
Speakers were asked to indicate their level of
agreement with statements such as “I identify with an
English/Mandarin-speaking culture” on a 7-point
scale (four statements per language). The sum of
points for English-oriented statements and Mandarin-
oriented statements was obtained, and a
Mandarin/English attitude ratio was calculated. A
ratio of 1 indicated that the speaker valued their
Mandarin and English-speaking identities equally,
while a ratio lower than one indicated a more English-
oriented attitude and a ratio greater than 1 suggested
a more Mandarin-oriented attitude (see [7] for details
of the original study).
For the current study, sound clips were extracted
from approximately the middle of the recording for
each condition of the direction-giving task. Each clip
was approximately 10 seconds in length and
contained a full inflection phrase (e.g. “into the
garage,” “now to the zoo”). Clips were carefully
selected to ensure that they did not contain
disfluencies, prominent dialect variation or feedback
from the confederate. Clips were extracted and
normalized for loudness using Praat [4]. This
procedure was followed for all recordings, resulting
in 24 clips in each of the three conditions. Each set of
24 clips was divided into three blocks consisting of 8
clips each. Block 1 contained clips from speakers 1-
8, block 2 contained clips from speakers 9-16, and
block 3 contained clips from speakers 17-24. Blocks
were arranged into six lists of 24 items such that two
blocks from the same condition never appeared on the
same list. The order of block selection was
counterbalanced across lists. Each listener rated only
one of the six lists. As a result, each rater judged all
speakers and all conditions but never rated a single
speaker more than once.
2.3. Procedures
Participants took the survey online via Qualtrics.
They rated 24 clips of non-native speech using three
7-point scales (7 being the highest). The prompts and
their scales were as follows: “How well did you
understand this person speaking English?” (e.g., “Not
well at all”; “Very poorly”; “Poorly”; “Moderately”,
“Fairly well”, “Well”, “Very well”), “Please rate the
strength of this person’s foreign accent, if any, when
speaking English” (“non-existent”; “Very weak”;
“Weak”; “Moderate”; “Fairly strong”; “Strong”;
“Very strong”) and “Please evaluate this person’s
overall proficiency in English” (“Not high at all”;
“Very poor”; “Poor”; “Moderate”; “Good”; “Near-
native”; “Native-like”). Participants were randomly
assigned to one of six lists of 24 clips. Clips were
presented in random order, and each list was rated an
equal number of times.
3. RESULTS
3.1. Overall variability
Figure 1 demonstrates the overall variability in rating
scores and the amount of spread across speakers in
terms of their perceived intelligibility, proficiency,
and accentedness. It shows that overall speakers were
perceived as fairly intelligible, with intermediate-to
high English proficiency (all scores are above the
mid-scale ‘moderate’ point for both attributes). All
participants’ speech was perceived as accented, rising
almost to the 6th point on the scale, labelled as
‘strong’.
Figure 1: Average accentedness, intelligibility and
proficiency ratings per speaker (speakers are
arranged in the order of decreasing intelligibility).
2234
Within these ranges there was a fair amount of
variability in all three dimensions, from almost
perfect intelligibility (6.58 on a 7-point scale, speaker
11), near-native proficiency (6 on a 7-point scale,
speaker 11), and weak accent (2.97, speaker 11) to
moderate intelligibility (4.28, speaker 22), moderate
proficiency (3.94, speaker 22) and strong accent
(5.59, speaker 18).
Intelligibility and proficiency ratings appear to be
in an almost perfect positive correlation with each
other (in fact, the correlation was significant: r[72] =
0.94, p < .001), while both are in a negative
relationship with perceived accentedness (r[72] = -
.876, p < .001 and r[72] = -.863, p < .001). The
correlation between accentedness and the other two
attributes is less consistent, indicating that this
dimension exhibits a greater degree of independence
from the other two.
3.2. Language attitudes
For the majority of the speakers in the sample (16 out
of 24) the Mandarin/English attitudes ratio was above
1, indicating a more positive attitude towards
Mandarin than English, but there was a fair amount
of variability, the ratio ranging from 0.53 to 1.71 (Fig.
2).
Figure 2: Average Mandarin/English ratio across
speakers.
3.2. Proficiency, intelligibility, and accentedness
ratings across interlocutor conditions
On average, speakers were rated the most proficient
and intelligible when addressing the Russian
interlocutor (Fig. 3), whereas the differences between
Mandarin and English conditions were less
pronounced and less consistent. Speech addressed to
the Mandarin interlocutor was also rated as the most
accented, while speech addressed to the Russian
interlocutor was rated as the least accented.
Figure 3: Average attribute scores by interlocutor
condition (L1 background).
3.3. Linear mixed model results
Since intelligibility, proficiency, and accentedness
scores were correlated with each other, a single global
evaluation score was created by taking an average of
the three (accentedness scale was reversed prior to
averaging). The resulting global scores were normally
distributed (Shapiro-Wilk test, p = .866) therefore no
further transformation was applied. The global
evaluation score was submitted as a dependent
variable to a Linear Mixed Model with Interlocutor
condition as a fixed factor, Mandarin/English attitude
ratio as a covariate, the interaction between the two,
and a random intercept for speaker.
Figure 4: Correlation between Mandarin/English
ratio and global evaluation score.
2235
The results showed no significant effect of the
Interlocutor condition and no significant interaction
between Interlocutor and Mandarin/English attitudes
ratio. The attitudes covariate was a significant
predictor of global evaluation score (β = -0.919, SE =
0.434, t = -2.115, p = 0.041). A higher
Mandarin/English ratio was associated with a lower
evaluation score (Fig. 4), indicating that speakers who
were less positively oriented towards English
language and culture were perceived as less
intelligible, less proficient, and more accented by
native English-speaking raters. It is worth mentioning
that significant correlations were also observed
between each one of the evaluative dimensions and
attitudes ratio. Additionally, no correlations were
found between attitudes ratio and participants AOA,
LOR, years of L2 schooling, and onset of acquisition.
4. DISCUSSION
Average differences in evaluation score as function of
interlocutor condition suggest that non-native
speakers aimed for greater intelligibility when
addressing the native Russian speaker. This makes
intuitive sense, since the Russian interlocutor has
neither the benefit of being a native speaker of
English, nor the benefit of a shared L1 background
with the Mandarin speaker [2, 10, 11, 19]. Therefore,
from the point of view of the Mandarin speakers, the
Russian interlocutor is in a greater need of increased
intelligibility than other interlocutors.
However, these average differences must be
interpreted with great caution since they were not
statistically significant. On the other hand, the lack of
statistical significance could be due to
methodological choices. The sound samples were
extracted from a relatively arbitrary point in a
recording of a naturalistic conversational task, which
means that a large amount of irrelevant variability
was unavoidable. This variability has likely reduced
the sensitivity of the perceptual test. In addition, the
sound samples were rather short, and it is possible
that they didn’t provide the raters with enough
information to arrive at a reliable evaluation of
participants’ speech. Finally, findings in [7] were not
entirely compatible with the view that speakers
modified the acoustics of their speech in order to
increase intelligibility. Instead, affective factors
could play a role. Results suggested that participants
may have modified their speech to indicate a greater
level of engagement and a positive stance in the
conversation. If this was indeed the case, eliciting
positive affect judgements of the speech samples
instead of intelligibility judgements would be a better
way to reveal interlocutor condition differences.
The only statistically significant effect detected in
the present data was that of a covariation between
participants attitudes ratio and their evaluation score.
A more positive attitude towards English was
associated with greater intelligibility, greater
proficiency, and lower accentedness, as indicated by
the global evaluation score. While it is unknown
whether a more positive L2 attitude leads to greater
success in L2 acquisition or whether greater prowess
at language learning ultimately results in a more
positive attitude, it is clear that the two are connected.
Moreover, this link was established on the basis of
evaluation of very brief speech samples, containing a
single inflectional phrase. Among other things, this
indicates that listeners are able to form such
judgements quickly and with limited information.
While it is certainly possible that the relationship
between language attitudes and perceived
intelligibility or accentedness of L2 is indirect and is
instead mediated by factors such as LOR, our data did
not reveal any significant correlations between
variables quantifying the onset and duration of L2
acquisition and attitudes ratio. Admittedly, our
sample of university students was quite homogeneous
with respect to these background characteristics.
Further research is necessary to establish whether
these or other circumstances of L2 acquisition could
mediate the relationship between language attitudes
and L2 speech intelligibility, accentedness, and
proficiency.
The ratings of intelligibility, proficiency, and
accentedness could not be analysed separately in the
present study because of covariation among them.
However, the three concepts are, in principle,
independent. That is, a speaker can be relatively
accented but highly intelligible [5, 17, 18]. Similarly,
high overall proficiency does not necessarily
guarantee high intelligibility [5]. Covariation among
these attributes in the present study could be
attributed to the fact that all three were evaluated
simultaneously for a given speaker. Blocking the
experiment by task rather than by speaker could de-
correlate the three dimensions, potentially revealing
different patterns of results.
To conclude, results of the study suggest that
language attitudes play an important role in
acquisition of second language speech. Acoustic
differences in non-native speech that are associated
with differences in language attitudes are
perceptually salient and detectable by native listeners.
From the pedagogical perspective, this study
confirms that students’ positive attitudes towards the
language can lead to improved intelligibility,
ultimately justifying work that language teachers do
to ensure that students have a positive orientation
towards the language and associated culture.
2236
5. REFERENCES
[1] Anderson, A., Bader, M., Bard, E., Boyle, E., Doherty,
G.M., Garrod, S., Isard, S., Kowtko, J., McAllister, J.,
Miller, J., Sotillo, C., Thompson, H.S., Weinert, R. 1991.
The HCRC Map Task Corpus. Language and Speech, I34,
351-366.
[2] Bent, T., Bradlow, A.R. The interlanguage speech
intelligibility benefit. J. Acoust. Soc. Am, 114(3), 1600-
1610.
[3] Birdsong, D., Gertken, L.M., Amengual, M. 2012.
Bilingual Language Profile: An Easy-to-Use Instrument to
Assess Bilingualism. COERLL, University of Texas at
Austin. Web. 25 Mar.
[4] Boersma, P., Weenink, D. 2009. Praat: Doing phonetics
by computer (Version 6.0.36) [Computer program].
Amsterdam, The Netherlands: University of Amsterdam.
[5] Derwing, T., Munro, M. 1997. Accent, intelligibility,
and comprehensibility: Evidence from four L1s. Studies in
SLA, 19(1), 1-16.
[6] Dörnyei, Z. (2014). The psychology of the language
learner: Individual differences in second language
acquisition. Routledge.
[7] Dmitrieva, O., Law, W.L., Lin, M., Wang, Y., Conklin,
J., Kentner, A., 2015. Language attitudes and listener-
oriented properties in non-native speech. Non-Native
Speech: Production and Perception. Proc. 18th ICPhS
Glasgow.
[8] Elliott, A. R. 1995. Field independence/dependence,
hemispheric specialization and attitude in relation to
pronunciation accuracy in Spanish as a foreign language.
Modern Language Journal, 79, 356-371.
[9] Foote, J. A., Trofimovich, P. 2018. Is it because of my
language background? A study of language background
influence on comprehensibility judgements. The Canadian
Modern Language Review, 74(2), 253-278.
[10] Gardner, R., & Lambert, W. (1972). Attitudes and
motivation in second-language learning. Rowley, Mass.:
Newbury House.
[11] Harding, L. 2012. Accent, listening assessment and
the potential for a shared-L1 advantage: A DIF perspective.
Language Testing, 29(2), 163-180.
[12] Huensch, A., Thompson, A. 2017. Contextualizing
attitudes toward pronunciation: Foreign language learners
in the United States. Foreign Language Annals, 50(2), 410-
432.
[13] Kissling, E. M. 2014. What predicts the effectiveness
of foreign-language pronunciation instruction?
Investigating the role of perception and other individual
differences. Canadian Modern Language Review, 70(4),
532-558.
[14] Levis, J. 2015. Learners’ views of social issues in
pronunciation learning. Journal of Academic Language
and Learning, 9, A42-A55.
[15] Lord, G. 2008. Podcasting communities and second
language pronunciation. Foreign Language Annals, 41,
364-379.
[16] Moyer, A. 1999. Ultimate attainment in L2
phonology: The critical factors of age, motivation and
instruction. Studies in SLA, 21, 81-108.
[17] Munro, M.J., Dewing, T.M. 1995a. Foreign accent,
comprehensibility, and intelligibility in the speech of
second language learners. Language Learning, 45(1), 73-
97.
[18] Munro, M.J., Dewing, T.M. 1995b. Processing time,
accent, and comprehensibility in the perception of native
and foreign-accented speech. Language and Speech, 38(3),
289-306.
[19] Munro, M.J., Dewing, T.M., Morton, S.L. 2006. The
mutual intelligibility of L2 speech. Studies in SLA, 28(1),
111-131.
[20] Shively, R. L. 2008. L2 acquisition of [β], [ð], and [ɣ]
in Spanish: Impact of experience, linguistic environment,
and leaner variables. Southwest Journal of Linguistics. 27,
79-115.
[21] Shuy, R. W., & Fasold, R. W. (1973). Language
attitudes: current trends and prospects.
2237
ResearchGate has not been able to resolve any citations for this publication.
Conference Paper
Full-text available
The study examines a number of acoustic properties of non-native speech directed to a native speaker, a non-native speaker with a shared first language background, and a non-native speaker with a different first language. Results demonstrate that the interlocutor condition interacts with the language attitudes factor: Participants with more positive attitudes towards their second language (English) differ along several acoustic dimensions from participants with more positive attitudes towards their first language (Mandarin), especially when interacting with native speakers of English. Expanded vowel space, higher articulation rate, and increased pitch adopted by English-oriented participants in interactions with native speakers of English may be indicative of their greater positive emotional involvement in the interaction. .
Article
Full-text available
The abstract for this document is available on CSA Illumina.To view the Abstract, click the Abstract button above the document title.
Article
Full-text available
Success in L2 pronunciation learning is affected by both individual differences and social influences on learning. While individual differences have been extensively researched, social influences have not. This study examines the beliefs and attitudes of advanced learners of English in regard to their pronunciation abilities and improvement. Twelve graduate students took part in four weeks of individualized pronunciation tutoring followed by interviews asking about their pronunciation, use of English, and their pronunciation in social contexts. The interviews revealed four images of their pronunciation learning. The first was that their spoken language skills left them feeling pulled in conflicting directions; the second was that they believed that accents could be 'caught' (like a cold) from the models around them (whether those models were seen as good or bad); the third concerned the students' views of accent and identity, which by and large were not seen as connected; and the fourth suggested that they saw themselves as separate from regular social contact in the L2. Each of these images involved contradictory beliefs about the nature of pronunciation improvement and its relationship to social interaction. These beliefs made improvement in pronunciation difficult. It is only by helping learners address these contradictory beliefs that greater pronunciation improvement will be possible.
Article
Full-text available
This study investigated second language (L2) learners’ perception of L2 sounds as an individual difference that predicted their improvement in pronunciation after receiving instruction. Learners were given explicit pronunciation instruction in a series of modules added to their Spanish as a foreign language curriculum and were then tested on their pronunciation accuracy. Their perception of the target sounds was measured with an AX discrimination task. Though the best predictor of pronunciation post-test score was pre-test score, perception made a unique and significant contribution. The other factors associated with better pronunciation of some L2 sounds were age, attitude, and time spent using Spanish outside the classroom. The results suggest that instructors should give adequate time for learners to hone their perception of target sounds at the outset of pronunciation instruction, because their initial ability to perceive the target sounds will, in part, determine how much they learn from such instruction. The results support models of L2 speech acquisition that claim that target-like perception is a precursor to target-like production, in this case in a formal learning context.
Article
While previous work has shown a relationship between pronunciation attitudes and pronunciation performance, the connection between language learning motivation and pronunciation attitudes has been underexplored. This study investigated the relationship between 195 foreign language learners’ attitudes toward pronunciation, the foreign languages studied, extramural language activity, and motivation. With an online three-part survey, information was collected about their language learning backgrounds; their ideal, ought-to, and anti-ought-to selves (Dörnyei, 2009; Thompson & Vásquez, 2015); and their attitudes toward pronunciation (Elliott, 1995a). The results from an exploratory factor analysis on the pronunciation items indicated a three-factor solution: lack of native speaker bias, importance of improving pronunciation, and importance of communication/skills other than pronunciation. There were significant group differences for class level and extramural language activity regarding learners’ perceived importance of improving pronunciation: Learners in upper-level classes as well as learners who participated in extramural language activity placed a higher importance on improving pronunciation. The results also indicated positive relationships between positive attitudes toward pronunciation and the ideal self, and between the desire to improve pronunciation and the anti-ought-to self. The findings have implications for the selection of materials and instructional approaches. © 2017 by American Council on the Teaching of Foreign Languages
Article
The abstract for this document is available on CSA Illumina.To view the Abstract, click the Abstract button above the document title.
Article
Research results over the past decades have consistently demonstrated that a key reason why many second language learners fail--while some learners do better with less effort--lies in various learner attributes such as personality traits, motivation, or language aptitude. In psychology, these attributes have traditionally been called "individual differences." The scope of individual learner differences is broad--ranging from creativity to learner styles and anxiety--yet there is no current, comprehensive, and unified volume that provides an overview of the considerable amount of research conducted on various language learner differences, until now.
Article
The abstract for this document is available on CSA Illumina.To view the Abstract, click the Abstract button above the document title.
Article
This paper reports on an investigation of the potential for a shared-L1 advantage on an academic English listening test featuring speakers with L2 accents. Two hundred and twelve second-language listeners (including 70 Mandarin Chinese L1 listeners and 60 Japanese L1 listeners) completed three versions of the University Test of English as a Second Language (UTESL) listening sub-test which featured an Australian English-accented speaker, a Japanese-accented speaker and a Mandarin Chinese-accented speaker. Differential item functioning (DIF) analyses were conducted on data from the tests which featured L2-accented speakers using two methods of DIF detection - the standardization procedure and the Mantel-Haenszel procedure - with candidates matched for ability on the test featuring the Australian English-accented speaker. Findings showed that Japanese L1 listeners were advantaged on a small number of items on the test featuring the Japanese-accented speaker, but these were balanced by items which favoured non-Japanese L1 listeners. By contrast, Mandarin Chinese L1 listeners were clearly advantaged across several items on the test featuring a Mandarin Chinese L1 speaker. The implications of these findings for claims of bias are discussed with reference to the role of speaker accent in the listening construct.