Content uploaded by Antonia Götz
Author content
All content in this area was uploaded by Antonia Götz on Aug 23, 2023
Content may be subject to copyright.
NON-NATIVE TONE PERCEPTION – WHEN MUSIC OUTWEIGHS
LANGUAGE EXPERIENCE
Antonia Götz 1,2 & Liquan Liu 1,3,4,5
1 The MARCS Institute for Brain, Behaviour and Development, Western Sydney University, Sydney, Australia,
2 Department of Linguistics, University of Potsdam, Potsdam, Germany,
3 School of Psychology, Western Sydney University, Sydney, Australia,
4 Center of Multilingualism across the Lifespan, University of Oslo, Oslo, Norway,
5 Centre of Excellence for the Dynamics of Language, Australian Research Council, Canberra, Australia.
ABSTRACT
This study examined how language (e.g.,
bilingualism, L2) and music (e.g., years of practising)
experiences improve lexical tone perception. A
substantial number of 532 participants from L1
Mandarin, L1 non-tone, bilingual L1 non-tone & L2
non-tone, and bilingual L1 non-tone & L2 tone
backgrounds were tested on their discrimination of
Mandarin tones. Results revealed that neither
bilingual nor second (tone or non-tone) language
experience affects novel tone perception. However,
listeners’ years of music training predicted perception
outcomes regardless of listeners’ language
backgrounds. These results indicate 1) learning a tone
language as L2 does not guarantee perceptual
advantage of non-native tones, even after years of
learning; 2) a myth of “bilingual advantage” in tone
perception, challenging the bilingual enhanced
acoustic sensitivity hypothesis in the perceptual
domain; and 3) learning a musical instrument helps
with tone perception across language groups,
exhibiting a cross-domain effect in the processing of
linguistic and musical pitch.
Keywords: tone perception, bilingualism, second
language perception, years of musical training, cross-
domain effect.
1. INTRODUCTION
In tone languages, tones distinguish word meanings.
Around 60-70% of the world’s languages are tonal
[1], and more than half of the world population speak
a tone language [2]. Most tone perception studies
have been conducted on individuals from tonal
languages, such as Mandarin or Cantonese, and there
is a limited understanding of how individuals from
non-tonal languages perceive tone and how second
language learning assist in perceiving lexical tones.
This project extended from existing tone perception
studies and investigated whether tone experience
learned from one’s second language, sequential
bilingual experience and music experience play a role
in tone perception. Specifically, the study can inform
the development of more effective language learning
strategies and interventions for individuals with tone
perception difficulties, as well as shed light on the
cognitive benefits of bilingualism and musical
training.
Figure 1: In [6], participants from five language
backgrounds (Australian English, Cantonese, Chinese
Mandarin, Singaporean Mandarin, Thai) unanimously
show the same pattern when perceiving static and dynamic
tones across four tone languages (Cantonese, Chinese
Mandarin, Singaporean Mandarin, Thai). A salience
hierarchy is hypothesised marking which type of tone
contrast is easier (higher) to discriminate than others
(lower) regardless of listeners’ language backgrounds.
Some tones or tone contrasts are naturally more
discriminable than others. In Mandarin tones, T1-T3
(level-dipping) appears more salient than T2-T3
(rising-dipping) for tone and non-tone language
speakers [3], and T2-T4 (rising-falling) exceeds other
contrasts in discriminability [4]. In Cantonese tones,
T2-T6 (high rising-low rising) leads to lower
discrimination accuracy than other contrasts across
listeners from tone and non-tone, monolingual and
bilingual backgrounds [5]. A recent collaborative
study involving participants from five tone and non-
tone language backgrounds reports a salience
hierarchy regarding the type of tone contrast (Figure
1). Prior research has established that tone language
speakers perceive tones more categorically than their
non-tone language peers [7]. What remains less clear
is whether tone language speakers perceive non-
native tones as well as native ones, and to what extent
listeners’ L2 tonal experience would lead to
successful tone discrimination. One recent study
testing Cantonese, Chinese, Singaporean and Thai
listeners’ perception of tones from their own versus
other languages reveals that tone language speakers
do not perceive non-native tones as good as their
native ones [6]. When testing Mandarin perception by
L2 Mandarin learners from L1 Cantonese or English
backgrounds, the two groups do not differ in their
overall accuracy on Mandarin tones, and both groups
show difficulty with the acoustically most difficult
T2-T3 tone pair in Mandarin [8]. Tone perception
appears to be difficult even for advanced L2 learners
[9], and listeners’ L1 knowledge and the acoustic
features of L2 tones both interfere with tone
perception.
With respect to bilingualism, simultaneous
bilingual experience has been shown to strengthen
tone perception in infancy [10], [11] and adulthood
[12] even when tone is not part of the language
repertoires of the bilingual speakers. Such perceptual
advantage is often attributed to the bilingual
environment typically more complex than a
monolingual one. The need of establishing two
phonological systems may well enhance listeners’
sensitivity to a third. Whether such advantage extends
to the cognitive domain is under heated debate [13].
As speakers’ L2 experience can also be considered as
a sequential bilingual experience, it is interesting to
see whether (non-tone) L2 experience may play a
role, as well as whether the length of L2 learning may
be a relevant factor in this case.
In terms of music experience, music training has
been shown to improve non-tone language speakers’
tone perception [14]–[16]. Listeners with no prior
tone language experience discriminate tones more
accurately when they are more musically trained [17],
[18]. Some studies suggest that non-tone language
speakers perceive (linguistic) tones in the same
fashion of musical tunes, as their performances of
pitch in language and music correlate [19]. These
findings point to a domain-general effect showing
that music experience can enhance non-tone language
speakers’ tone perception. Bilinguals and tone
language speakers do not show such correlation [12],
suggesting the interference of bilingual and L1
experience on tone perception. The research
questions of the current study are: 1. Do different tone
contrasts lead to different perceptual outcomes? 2.
Does (L1 or L2) tone language or sequential
bilingualism experience strengthen novel tone
perception? Do listeners’ years of L2 experience play
a role? 3 Does music experience (years of music
training) modulate tone perception? We predicted
that based on the salience hierarchy [6], Mandarin T2-
T4 (dynamic different) would be the most salient
contrast, followed by T1-T2/T3/T4 (static-dynamic),
and then T3-T2/T4 (dynamic similar). Moreover,
listeners’ L2 (tone, non-tone) and music experiences
facilitate tone perception, and the magnitude of
facilitation is directly relevant to their years of
experience.
2. METHODS
2.1. Participants
A total number of 814 subjects participated in the
study. Of these participants, 191 did not complete the
online experiment and another 91 participants did not
provide valid consent sharing data for analysis. The
final sample included 532 participants from multiple
language backgrounds who were divided into three
language groups (Table 1).
Group
N
Mage
(range)
Years of
L2 (mean
(SD))
Years
of
music
(mean
(SD)
L1 Mandarin
25
18-25
12.65
(4.55)
2.93
(2.91)
Mono (non-
tone L1)
44
18-25
0.48
(0.09)
2.83
(2.93)
Bi (non-tone
L1 + non-
tone L2)
429
18-25
8.97
(5.14)
3.26
(3.53)
Bi (non-tone
L1 + tone
L2)
34
18-25
9.85
(4.94)
3.09
(3.05)
Table 1: Participants were categorised into three
groups. Here, Mono refers to monolinguals and Bi
refers to sequential bilinguals.L1 Mandarin speakers
had L2 experience with non-tone languages but not in
another tone langauge. Years of music refers to the
mean and standard deviation of years of musical
training that was self-indicated by the participants.
2.2. Stimuli
The stimuli consisted of 12 monosyllabic Mandarin
non-words (/tou/, /bou/, /ɕye/, /pye/, /pian/, /fian/, /jy/,
/ty/, /bi/, /gi/, /gua/, /lua/) with legal phonotactic
structures. Each syllable was produced with the four
Mandarin tones (T1, T2, T3 and T4). The length of
each syllable was 250 ms. The final stimulus set
consisted of 72 stimuli: 12 syllables x 6 tone contrasts
(T1-T2, T1-T3, T1-T4, T2-T3, T2-T4, T3-T4). A
Mandarin native speaker produced six tokens of each
syllable and we included two of the tokens into the
experiment. By including different tokens, we
prevent that the participants make their decision
based on acoustic information alone. The same token
was never repeated within one trial of the task. For
example, in AAB trials, if token 1 was used as the A
sound, token 2 would appear as the sound for X (the
second A. In addition to the variability of the different
tokens, the same word was never repeated within one
Tone Contrast. All stimuli were normalized in
intensity (70 dB).
2.3. Procedure
The experiment ran online by using the online
platform Labvanced [20] at a quiet place and
participants were asked to wear headphones. Each
experiment started with two practice trials to
familiarise the participants with the AXB
discrimination task. Participants were asked to press
a key as accurate and quick as possible if the second
syllable was more similar to the first one (AAB, via
key 1) or the third (ABB, via key 3) syllable. All
contrast were also presented in the reverse order to
counterbalance participants’ responses. In each trial,
the A and B sounds consisted of a different tone
category. The interstimulus interval was 1000 ms and
the intertrial interval was 3000 ms. The time-out of
the response time was set to 2500 ms, measured at the
end of the third syllable. Trials with responses after
the time-out were not repeated. The next trial started
immediately following the previous trial and trials
were randomised across participants. A break was
included after 25%, 50% and 75% of the experimental
trials and its length was participant controlled;
participants continued the experiment by pressing a
key. No feedback was provided to the participants.
3. RESULTS
All statistical analyses were performed by using R
[21] and the lme4 package [22]. Plots were generated
by using ggplot2 [23]. General Linear Mixed Effects
regression models were constructed with the maximal
random and fixed factor structure with accuracy
(binomial answer as 1 or 0) as dependent variable.
Language background was coded as the comparison
between L1 native Mandarin (coded as 0.5) and
monolingual L1 (coded as -0.5), monolingual L1
(coded as 0.5) and bilingual L2 Non-tone speakers
(coded as -0.5) and bilingual L2 Non-tone speakers
(coded as 0.5) versus L2 Tone speakers (coded as -
0.5). Within the Tone Contrast discrimination, we
predicted the following hierarchy: T2-T4 > T1-
T2/T3/T4 > T3-T2/T4. Following this hierarchy, we
applied the subsequent contrast comparisons: T2-T4
was compared to the mean of T1-T2, T1-T3 and
T1T4, T3-T4 was compared to the mean of T1-T2,
T1-T3 and T1T4, T3-T4, T3-T4 was compared to T2-
T3, T1-T3 was compared to T1-T4 and T1-T2 was
compared to T1-T3. Listener’s discrimination of the
Tone Contrast showed the expected patterns: T2-T4
yielded higher accuracy than T1-TN contrasts (β(SE)
= 0.461 (0.019), z = 23.289, p < .001), which yielded
higher accuracies than T3-T4 (β(SE) = 0.165(0.014),
z = 11.957, p < .001), which was further higher than
T2-T3 (β(SE) = 0.254(0.024), z = 10.390, p < .001).
Our results (see also Figure 2) revealed that
L1 Mandarin speakers are not better in discriminating
Tone Contrasts than L1 Non-tone speakers (β(SE) = -
0.031(0.065), z = 0.478, p = 0.633). L2 tone language
speakers are not better in discriminating Tone
Contrasts than L2 Non-tone language speakers (β(SE)
= -0.043(0.041), z = -1.054, p = 0.291) and the
listener’s years of experience did not interact (β(SE)
= 0.035(0.070), z = 0.498, p = 0.619). Similarly, L2
listeners of a non-tone language were not better in
discriminating the Tone Contrasts than L1 Non-tone
language listeners (β(SE) = -0.020(0.034), z = -0.594,
p = 0.553), and duration of the second language did
not interact (β(SE) = -0.034(0.070), z = 0.483, p =
0.629). However, music experience modulated Tone
Contrast discrimination. The more music experience
listeners have, the better their tone discrimination
(β(SE) = 0.011(0.005), z = 2.077, p = 0.0383).
Figure 2: Accuracy results from the speech
discrimination task separated by the four groups of L1
and L2 tone language experience (Mandarin, Monolingual
L1 Non-tone, Bilingual L2 Non-tone, Bilingual L2 Tone)
4. DISCUSSION
The current study investigated the extent to which
L2 (tone or non-tone) language and music
experiences modulate tone perception. With respect
to tone contrasts, a salience hierarchy as predicted
was reported. Contrasting our predictions, listeners’
L1 and L2 language experience does not alter tone
perception, regardless of whether the language is
tonal or how many years they have learned the
language. However, music experience plays an
important role for speakers across language
backgrounds, with positive relationship between
years of music training and successful tone perception
outcomes.
In terms of Mandarin tone contrasts, dynamic
tones with different pitch directions (T2-T4, rising-
falling) were the easiest to discriminate, followed by
static-dynamic tone contrasts (T1-T2/T3/T4, flat-
contour tones) and then dynamic tones with similar
pitch patterns (T3-T2/T4, dipping-rising/falling). Our
findings replicate the salience hierarchy reported in
Liu and colleagues [6], suggesting that tone acoustics
is critical for listeners’ tone perception across
language backgrounds.
Learning a tone language as L2 does not lead to
successful tone perception. Previous studies have
reported similar findings for L1 speakers of a non-
tone and even a tone language learning another tone
language as L2 [8]. Even advanced L2 learners may
find it difficult to (re-)establish a tone category [9],
conforming to our results that years of L2 experience
does not affect tone perception outcomes. The finding
is in line with another study on segmental contrast
perception. Catalan has an /e/–/ɛ/ contrast while
Spanish has only one /e/ closer to and more open than
the Catalan /e/. Pallier and colleagues [24] examined
two types of Catalan-Spanish bilinguals’ perception
of the Catalan contrast. The Catalan-dominant
bilinguals are exposed to Catalan since birth, whereas
the Spanish-dominant bilinguals are exposed to
Spanish first and get in touch with Catalan after 6
years of age when they start in kindergarten or
primary school. Only Catalan-dominant but not
Spanish-dominant bilinguals discriminate the Catalan
contrast. Results of the vowel perception among
Spanish-dominant bilinguals are in accordance with
the findings of tone perception in the current study.
The overall findings also demonstrate the importance
of early exposure on phonological category
establishment and later perception, mirroring studies
reporting perceptual advantages [25], [26] and neural
traces [27] of phonology of the birth language among
adoptees who are adopted to a new language
environment as early as 6 months after birth.
Although a simultaneous bilingual experience has
been shown to facilitate tone perception [10], a
sequential one does not appear to reach a similar
effect. While Dutch sequential bilinguals show
reduced tone discrimination abilities than their
Mandarin peers, Dutch simultaneous bilinguals stand
in the middle between these two groups in their
performances [12]. The overall findings add to the
limited data of how sequential bilingual may impact
speech perception, suggesting the importance of early
linguistic diversity on tone perception. The lack of
significant results between the language groups may
be attributed to the differences in group sizes and the
resulting larger variability in Bilingual L2 Non-tone
group.
Regarding music experience, the current results
conform to and extend on prior studies [14]–[16].
That is, listeners do not have to be musicians to show
an advantage in tone perception. Their years of music
training is directly relevant to tone perception
irrespective of their language backgrounds. The
finding that overall music experience can facilitate
speech perception provides new insights into
potential cross-domain strategies in L2 language
learning, especially when the target language is tonal.
5. CONCLUSION
When perceiving non-native tones, listeners’ music
experience plays an important role. Its effect may be
more evident than L2 experiences, as neither L2 tone
language nor sequential bilingual experience appears
to affect perception of foreign tones. The overall
findings provide implications for early, consistent
diversity in and exposure to language and music.
6. REFERENCES
[1] M. Yip, Tone. Cambridge University Press, 2002.
[2] V. A. Fromkin, Tone: A linguistic survey.
Academic Press, 2014.
[3] Z. Zeng, L. Liu, A. Tuninetti, V. Peter, F.-M. Tsao,
and K. Mattock, “English and Mandarin native
speakers’ cue-weighting of lexical stress: Results
from MMN and LDN,” Brain and Language, vol.
232, p. 105151, 2022.
[4] T. Huang and K. Johnson, “Language specificity in
speech perception: Perception of Mandarin tones by
native and nonnative listeners,” Phonetica, vol. 67,
no. 4, pp. 243–267, 2010.
[5] X. Tong, S. M. K. Lee, M. M. L. Lee, and D.
Burnham, “A tale of two features: Perception of
Cantonese lexical tone and English lexical stress in
Cantonese-English bilinguals,” PloS one, vol. 10,
no. 11, p. e0142896, 2015.
[6] L. Liu et al., “The tone atlas of perceptual
discriminability and perceptual distance: Four tone
languages and five language groups,” Brain and
Language, vol. 229, p. 105106, 2022.
[7] P. A. Hallé, Y.-C. Chang, and C. T. Best,
“Identification and discrimination of Mandarin
Chinese tones by Mandarin Chinese vs. French
listeners,” Journal of Phonetics, vol. 32, no. 3, pp.
395–421, 2004.
[8] Y.-C. Hao, “Second language acquisition of
Mandarin Chinese tones by tonal and non-tonal
language speakers,” Journal of Phonetics, vol. 40,
no. 2, pp. 269–279, Mar. 2012, doi:
10.1016/j.wocn.2011.11.001.
[9] E. Pelzl, E. F. Lau, T. Guo, and R. DeKeyser,
“Advanced second language learners' perception of
lexical tone contrasts,” Studies in Second Language
Acquisition, vol. 41, no. 1, pp. 59–86, 2019.
[10] L. A. Petitto, M. S. Berens, I. Kovelman, M. H.
Dubins, K. Jasinska, and M. Shalinsky, “The
‘Perceptual Wedge’ hypothesis as the basis for
bilingual babies’ phonetic processing advantage:
New insights from fNIRS brain imaging,” Brain
Lang, vol. 121, no. 2, pp. 130–143, May 2012, doi:
10.1016/j.bandl.2011.05.003.
[11] L. Liu and R. Kager, “Perception of tones by
bilingual infants learning non-tone languages,”
Bilingualism: Language and Cognition, vol. 20, no.
3, pp. 561–575, 2017.
[12] L. Liu, A. Chen, and R. Kager, “Simultaneous
bilinguals who do not speak a tone language show
enhancement in pitch sensitivity but not in
executive function,” Linguistic Approaches to
Bilingualism, vol. 12, no. 3, pp. 310–346, 2022.
[13] M. Lehtonen, A. Soveri, A. Laine, J. Järvenpää, A.
De Bruin, and J. Antfolk, “Is bilingualism
associated with enhanced executive functioning in
adults? A meta-analytic review.,” Psychological
bulletin, vol. 144, no. 4, p. 394, 2018.
[14] D. Burnham, R. Brooker, and A. Reid, “The effects
of absolute pitch ability and musical training on
lexical tone perception,” Psychology of Music, vol.
43, no. 6, pp. 881–897, Nov. 2015, doi:
10.1177/0305735614546359.
[15] A. Cooper and Y. Wang, “The influence of
linguistic and musical experience on Cantonese
word learning,” The Journal of the Acoustical
Society of America, vol. 131, no. 6, pp. 4756–4769,
2012.
[16] P. C. Wong and T. K. Perrachione, “Learning pitch
patterns in lexical identification by native English-
speaking adults,” Applied Psycholinguistics, vol.
28, no. 04, pp. 565–585, 2007.
[17] J. A. Alexander, P. C. M. Wong, and A. Bradlow,
“Lexical tone perception in musicians and non-
musicians,” presented at the European Conference
on Speech Communication and Technology,
Lissabon, Portugal, 2005.
[18] B. Chandrasekaran, A. Krishnan, and J. T.
Gandour, “Relative influence of musical and
linguistic experience on early cortical processing of
pitch contours,” Brain Lang, vol. 108, no. 1, pp. 1–
9, Jan. 2009, doi: 10.1016/j.bandl.2008.02.001.
[19] A. Chen, L. Liu, and R. Kager, “Cross-domain
correlation in pitch perception, the influence of
native language,” Language, Cognition and
Neuroscience, vol. 31, no. 6, pp. 751–760, 2016.
[20] H. Finger, C. Goeke, D. Diekamp, K. Standvoß,
and P. König, “LabVanced: a unified JavaScript
framework for online studies,” in International
conference on computational social science
(cologne), 2017.
[21] R Core Team, “R: A language and environment for
statistical computing. R Foundation for Statistical
Computing.” Vienna, Austria., 2021. [Online].
Available: https://www.R-project.org/
[22] D. M. Bates, Maechler, M., Bolker, B., and Walker,
S., “Fitting Linear Mixed-Effects Models using
lme4,” Journal of Statistical Software, 2015.
[23] H. Wickham, “Programming with ggplot2,” in
ggplot2: Elegant Graphics for Data Analysis, H.
Wickham, Ed. Cham: Springer International
Publishing, 2016, pp. 241–253. doi: 10.1007/978-3-
319-24277-4_12.
[24] C. Pallier, L. Bosch, and N. Sebastián-Gallés, “A
limit on behavioral plasticity in speech perception,”
Cognition, vol. 64, no. 3, pp. B9–B17, 1997.
[25] W. Choi, X. Tong, F. Gu, X. Tong, and L. Wong,
“On the early neural perceptual integrality of tones
and vowels,” Journal of Neurolinguistics, vol. 41,
pp. 11–23, 2017.
[26] W. Choi, X. Tong, and S. H. Deacon, “Double
dissociations in reading comprehension difficulties
among Chinese–English bilinguals and their
association with tone awareness,” Journal of
Research in Reading, vol. 40, no. 2, pp. 184–198,
2017.
[27] L. J. Pierce, J.-K. Chen, A. Delcenserie, F.
Genesee, and D. Klein, “Past experience shapes
ongoing neural patterns for language,” Nature
Communications, vol. 6, no. 1, pp. 1–11, 2015.
