TWO LANGUAGES, TWO PITCH RANGES: THE CASE OF JAPANESE-ENGLISH SEQUENTIAL BILINGUALS

Abstract

We designed an experiment aiming to investigate variation in the pitch range of the two languages of Japanese-English sequential bilinguals. Data were collected from eight Japanese-English bilinguals and eight English monolinguals in London (UK). Preliminary results from a reading task show significant differences across the pitch profiles in the English of the bilinguals and the English of the monolinguals, as well as significant cross-linguistic differences across the pitch profiles of the two languages of the bilinguals. The observed between-group differences are consistent with previous findings on pitch profiles of languages spoken as an L1 and an L2. The within-group differences show an unexpected pattern: the bilinguals' mean F0s in English are higher than their mean F0s in Japanese, irrespective of whether female or male. This could be attributed to increased stress due to L2 anxiety or to a potential change in the language norms of Japanese.

Full text

TWO LANGUAGES, TWO PITCH RANGES:
THE CASE OF JAPANESE-ENGLISH SEQUENTIAL BILINGUALS
Elisa Passoni, Esther de Leeuw, Erez Levon
Queen Mary University of London
{e.passoni, e.deleeuw, e.levon}@qmul.ac.uk
ABSTRACT
We designed an experiment aiming to investigate
variation in the pitch range of the two languages of
Japanese-English sequential bilinguals. Data were
collected from eight Japanese-English bilinguals and
eight English monolinguals in London (UK).
Preliminary results from a reading task show
significant differences across the pitch profiles in the
English of the bilinguals and the English of the
monolinguals, as well as significant cross-linguistic
differences across the pitch profiles of the two
languages of the bilinguals. The observed between-
group differences are consistent with previous
findings on pitch profiles of languages spoken as an
L1 and an L2. The within-group differences show an
unexpected pattern: the bilinguals’ mean F0s in
English are higher than their mean F0s in Japanese,
irrespective of whether female or male. This could be
attributed to increased stress due to L2 anxiety or to a
potential change in the language norms of Japanese.
Keywords: Pitch range, prosody, bilingualism, L2
acquisition, sex
1. INTRODUCTION
There is a substantial body of research on pitch which
has compared languages [4, 22, 27], dialects [15] and
social groups [6] showing that differences in pitch can
be due to (1) different phonological and intonational
structures, and/or (2) extra-linguistic factors,
including culture and emotional state. A few studies
have also compared F0 profiles in first (L1) and
second (L2) language speech [10, 33] showing that
pitch varies across the two languages of bilinguals
and this has been attributed to L1 prosodic transfer.
Following Ladd [16], pitch is considered a
manifestation of Fundamental Frequency Range
(FFR). There is a general consensus that pitch range
varies across two quasi-independent dimensions: (1)
level, which refers to the overall height of an
individual’s F0 and (2) span, which refers to the range
of F0s in a speech sample. However, as noted in [21],
there is no general consensus on the best approach for
quantifying these two dimensions. Here, we follow
methods operationalized in [27] for level and [17] for
span (see section 2.4 for details).
Previous research has reported that Japanese
females use a much higher pitch level than Euro-
American females [6, 19, 24, 25], due to the socio-
cultural constraints linked to ‘being female’ in Japan
[19, 24, 25]. In contrast, Japanese males have been
reported to use a low pitch level to maintain a “cool”
profile [19]. Using a reading task, Graham [12]
investigated cross-linguistic variation in the pitch of
simultaneous balanced Japanese-English bilinguals
and found that both females and males used a higher
pitch level and a wider pitch span in Japanese
compared to English. It was considered of interest to
investigate sequential bilinguals as they might be
more subject to linguistic and cultural transfer from
their L1 to their L2 [10], which could be modulated
by differences in Age of Acquisition (AoA), Length
of Residency (LoR) in the L2 country and L2
proficiency (as shown in [33]). The term bilingual is
used to describe people who use two or more
languages in their daily lives [13], in line with similar
research examining differences in individuals who
speak more than one language [18].
The present study is an initial stage investigation
on the pitch range of the two languages of Japanese-
English sequential bilinguals. To this end, we
compared the pitch range of the bilinguals in English
to the pitch range of native monolingual speakers of
English [as in 10, 22], as well as the pitch range of the
two languages of the bilinguals [as in 12, 19, 24, 25].
Our corpus was also balanced by sex, i.e. “the
individual biological status as female, male or
intersex” [3, p.2], allowing us to look into the effect
of this variable in the speech of the two groups of
participants. We expected to find differences in pitch
profiles between English as an L1 and an L2, as well
as a difference in the pitch range of the two languages
of the bilinguals. We also expected Japanese females
to show greater differences in their pitch profiles both
across sex groups and across languages, due to
Japanese language socio-cultural norms.
2. METHODS
This study was part of a larger study carried out both
in London (UK) and in Tokyo (JP).

2.1. Participants
Sixteen participants, all residents of London, UK,
took part in the study, divided into two groups: (1) 8
Japanese-English sequential bilinguals – JEB (4
females and 4 males) and (2) 8 SSBE functional
monolinguals – EM (4 females and 4 males).
Prior to data collection, participants were asked to
complete a language background questionnaire
adapted from LEAPQ [20] and MSI-Goldsmiths [23].
All bilingual participants considered Standard
Japanese to be their L1, followed by English, which
they all acquired from childhood through formal
education. Overall, the bilingual participants of this
study were consecutive bilinguals, varying in degrees
of AoA and LoR and self-reported L2 proficiency on
a scale from 0 (none) to 10 (perfect) (see Table 1
below). In contrast, the SSBE native speakers
reported English as their L1 and to have studied some
languages mainly at secondary school but not to be
proficient nor to use them on a daily basis. The 8 EMs
only ever resided in England and had only been
abroad for holidays.
Table 1: Language background information of
participants of both groups (SD in brackets).
Group
Age
(yrs)
AoA
(yrs)
LoR
(yrs)
L2
proficiency
JEB
30
(7.1)
11
(3.1)
6
(4.4)
7.5
(1.2)
EM
25 (6.4)
n/a
n/a
n/a
2.2. Stimuli
The stimuli which the participant read out consisted
of 16 English sentences and their translations in
Japanese taken from [12]. They comprised four types
of sentences: Alternative questions (Alt_QS),
Declarative questions (Dec_QS), Declarative
statements (Dec_ST) and Wh-questions (Wh_QS).
Four sentences for each type were used. These
sentences were chosen because they are favourable
for pitch analysis, i.e. they contained a high amount
of fully voiced segments, whilst being short enough
to typically correspond to a single intonational phrase
(IP) and so control for declination effects [28].
2.3. Experimental Procedure
Data collection took place in a soundproof IAC booth
in the QMUL Phonetics Lab. The recording chain was
a Røde NT1-A condenser microphone (cardioid polar
pattern) and a Steinberg UR22 audio interface
(microphone preamp and analogue to digital
converter). All audio was recorded on a MacBook Pro
at a sample rate of 44.1 kHz, 16-bit.
The reading task was the second task of a longer
study comprising 3 tasks and 3 questionnaires.
Participants were comfortably seated at a computer
and presented with an interface created in PsychoPy
1.85.2 [29]. Participants were instructed to read each
sentence naturally without changing the words in any
way. To minimize interference from the investigator
[11] and avoid influences in terms of phonetic
imitation [1] or gender interactions [7] on the speech
of the participants, instructions were given by a
gender-neutral animated character created in Adobe
Character Animator [2], who ‘spoke’ to the
participants via speech bubbles. The study was first
created in English and then translated into Japanese
by a professional translator. The translation was
subsequently blindly back-translated into English by
another translator, as recommended for cross-cultural
research [9].
Each stimulus was presented only once, in a
randomised order. Bilinguals did the study in both of
their languages separately with a 30-min break
between halves to account for language modes [14];
languages were counterbalanced across participants.
2.4. Phonetic analysis
FFR measurements were made for the two aspects of
pitch range described previously: level and span.
Recordings were first segmented and then visually
and auditorily inspected in Praat [8] before running a
customized script to extract pitch variables across
each sentence. Waveforms and spectrograms were
examined in 5-10 second intervals to check for octave
jumps and/or doubling, as well as for sections of
creaky voice which were removed from the analysis.
For females, the pitch floor was set at 100 Hz and
pitch ceiling at 500 Hz. For males, pitch floor was set
at 75 Hz and ceiling at 300 Hz. Following [27], for
pitch level, mean, min and max F0 were extracted; for
pitch span, the 80% range was obtained as in [17].
3. BETWEEN GROUP DIFFERENCES
We first aimed to expand upon prior work on L2
acquisition of prosody by asking whether JEB and
EM differed in the use of pitch range in English.
3.1. Statistical model
Linear mixed-effects models for measures of both
pitch level and span were built in R [30] using the
lme4 package [5]. Speaker and sentence were
included as random intercepts. Predictor variables
included Group (JEB vs EM), Sex (Female vs Male),
Sentence type (Alt_QS, Dec_QS, Dec_ST, Wh_QS),
and interactions between Group:Sex and
Group:Sentence type.

3.2. Results: Pitch level
Figure 1: Boxplot of mean F0 (Hz) by Sentence
type, divided by Group.
There was a significant effect of Group (ß=25, 3.610,
p=.002), i.e. Japanese-English bilinguals showed an
overall higher mean F0 in their English compared to
the SSBE monolinguals. There was also a significant
effect of Sentence type (ß=-16, -6.652, p<.0001), i.e.
Declarative statements elicited an overall lower mean
F0; as well as a significant effect of Sex (ß=-96, -
13.924, p< .0001), i.e. males showed an overall lower
mean F0 compared to females. There were no
significant interactions.
Due to clear differences in mean F0 across male
participants visible in Figure 1, data were further
partitioned by sex and two separate mixed effect
models were run for female and male participants
separately. Model parameters were the same as for the
previous analysis. Bonferroni-corrected results show
that Japanese males’ mean F0 was significantly
higher than the SBBE males (ß=32, 2.766, corrected
p =.024). Difference in mean F0 between Japanese
females’ and SBBE females did not reach
significance (ß=18, 2.751, p=.051).
Pearson product-moment correlations were run to
investigate the potential relationship between mean
F0 of the bilinguals and AoA, LoR and L2
Proficiency. No significant relationships were found.
3.3. Results: Pitch span
There was a significant effect of Group (ß=3.3, 3.121,
p=.011), i.e. Japanese-English bilinguals showed a
wider span in their English compared to the SSBE
monolinguals. There was also a significant effect of
Sentence type (ß=-1.9, -2.493, p<.0001), i.e.
Declarative questions elicited a narrower span.
Declarative statements also elicited an overall
significantly narrower span (ß=-2.1, -2.680,
p<.0001).
As for pitch level, separate models were run for
female and male participants. Male bilinguals showed
a significant wider span than male monolinguals
(ß=3.9, 3.249, corrected p=.011), whereas
differences in the female bilinguals did not reach
significance (ß=3.9, 2.481, p=.028).
Pearson product-moment correlations were run to
investigate the potential relationship between span of
the bilinguals and AoA, LoR and L2 Proficiency. No
significant relationships were found.
Figure 2: Boxplots of span (ST) by Sentence type,
divided by Group.
4. WITHIN GROUP DIFFERENCES
Having observed between-group differences, we next
investigated whether there was variation within the
bilinguals.
3.1. Statistical model
Linear mixed-effects models of both pitch level and
span were created as before. Speaker and sentence
were again included as random effects. Predictor
variables included Task language (English vs
Japanese), Sex, Type of sentence, and interactions
between Task language:Sex and Task
language:Sentence type.
3.2. Results: Pitch level
Figure 4: Boxplots of mean F0 (Hz) by Sentence
type, divided by Task language for bilinguals.
There was a highly significant effect of Task

language (ß=-23, -5.613, p<.0001), i.e. Japanese
evidenced an overall lower mean F0 compared to
English in the speech of the bilingual participants.
There was also a highly significant effect of Sex (ß=-
88, - 9.977, p<.0001), i.e. overall males’ mean F0
were significantly lower than the females’, but no
significant interaction between Task language and
Sex. Again, Declarative statements elicited a highly
significant lower mean F0 in both languages (ß=-19,
-4.172, p<.0001). There was also a highly significant
interaction between Task language and Sentence type
(ß=18, 3.409, p<.0001); i.e. in Japanese, Declarative
questions elicited a mean F0 higher than in English.
3.2. Results: Pitch span
Figure 5: Boxplots of span (ST) by Sentence type,
divided by Task language for bilinguals.
There was a highly significant main effect of Task
language (ß=2.3, 9.773, p<.0001), i.e. Japanese
evidenced an overall wider span compared to English
in the speech of the bilingual participants. No main
effect of Sex was found, nor interactions between
Task language and Sex. Declarative sentences elicited
an overall narrower span (ß=-1.1, -3.001, p=.009) and
Wh-questions an overall wider span (ß=0.8, 0.542,
p=.039).
4. DISCUSSION
Though preliminary, the results yielded some
noteworthy observations regarding the pitch profiles
of the bilinguals.
First, the study confirms and extends findings
about differences in pitch range between languages
spoken as an L1 and L2. We found that the bilinguals’
pitch range in English differed from the pitch range
of the monolinguals: specifically, the model estimate
for the bilinguals’ mean F0 in English was 25Hz
higher than the monolinguals’ mean F0. Moreover,
the model estimate for the bilinguals’ span was 3.3ST
wider than the monolinguals’ span. This is consistent
with previous research on the pitch range of L2
speakers [10, 33] and could be attributed to a transfer
from the L1 to the L2.
To investigate the possibility of such transfer, a
separate analysis compared the pitch ranges of the
bilinguals across their two languages. As expected,
we found a difference between the pitch profiles of
the bilinguals in their two languages; but
unexpectedly our results showed that the mean F0 of
the Japanese-English bilinguals in English was
significantly higher than their mean F0 in Japanese
for both women and men. This is not consistent with
mean F0 patterns previously reported for Japanese-
English bilinguals [12, 19, 24, 25]. Research on SLA
has shown that reading tasks correlate with increased
anxiety in students [32], and previous work on pitch
range and emotions has shown that stress and
fear/panic lead to higher mean F0 values [31]. This
may account for why our bilinguals had not only a
higher mean F0 in English than the monolinguals, but
also a higher mean F0 in their English than in their
Japanese. However, it may also be that language
norms in Japanese have changed since the
aforementioned studies were undertaken. Span results
are consistent with [12].
We were also interested in the effect of sex of
speaker on pitch range. Our results did not show sex
related differences, i.e. all bilinguals’ English mean
F0s were higher than the mean F0s of SBBE speakers
and all bilinguals’ mean F0s in English were higher
than their mean F0s in Japanese. Interestingly, if
anything, our results suggested that Japanese-English
males are the ones driving the difference in mean F0
across bilingual and monolingual groups. Previous
literature has reported an effect of gender on the mean
F0 of Japanese monolinguals and bilinguals, whereby
Japanese females’ high mean F0s were explained as a
consequence of gender-roles in the Japanese culture
[19, 24, 25, 26]. Why did Japanese-English males,
and not females, have a significantly higher mean F0
in English compared to monolinguals of the same sex
in this study? At present, it is difficult to draw any
definite conclusion due to the small sample size of the
current study, but since Japanese males have been
reported to use a low mean F0, which in English
implies ‘being cool…even unpleasant’ [19, p.83], it
might be argued that these speakers increased their
mean F0 in English, potentially subconsciously, to
ensure they portrayed themselves as pleasant.
In sum, our results showed an unexpected pattern
in the mean F0 of the two languages of Japanese-
English bilinguals in that the bilinguals’ pitch level in
English was significantly higher than their pitch level
in Japanese, irrespective of sex. Further investigation
using a larger data set, natural speech and/or different
methods is needed to deepen our understanding of the
potentially changing pitch patterns in this population.

5. REFERENCES
[1] Adank, P., Stewart, A. J., Connell, L., & Wood, J.
(2013). Accent imitation positively affects language
attitudes. Frontiers in Psychology, 4.
[2] Adobe. 2017. Adobe Character Animator CC (Version
Beta). Adobe Systems incorporated.
[3] American Psychological Association. (2012).
Guidelines for psychological practice with lesbian, gay,
and bisexual clients. American Psychologist, 67(1), 10–
42.
[4] Andreeva, B., Demenko, G., Wolska, M., Möbius, B.,
Zimmerer, F., Jügler, J., & Trouvain, J. 2014.
Comparison of pitch range and pitch variation in Slavic
and Germanic languages. Proceeding 8th Speech
Prosody, Dublin.
[5] Bates, D., Maechler M., Bolker B., Walker, S. 2015.
Fitting Linear Mixed-Effects Models Using lme4.
Journal of Statistical Software, 67(1), 1-48.
[6] van Bezooijen, R. 1995. Sociocultural aspects of pitch
differences between Japanese and Dutch women.
Language and Speech, 38 (Pt 3), 253–265.
[7] Biemans, M. (1998). The Effect of Biological Gender
(Sex) and Social Gender (Gender Identity) on Three
Pitch Measures. Linguistics in the Netherlands, 15, 41–
52.
[8] Boersma, P., & Weenink, D. 2016. Praat: doing
phonetics by computer. (Version 6.0.26). Retrieved
from http://www.praat.org/
[9] Brislin, R. W. 1970. Back-Translation for Cross-
Cultural Research. Journal of Cross-Cultural
Psychology, 1(3), 185–216.
[10] Busà, M. G., & Urbani, M. 2011. A cross linguistic
analysis of pitch range in English L1 and L2. In
Proceedings 17th ICPhS, Hong Kong, 380–383.
[11] Carrie, E. & Drummond, R. 2017. The Accent Van:
Methods in community-oriented linguistic research,
UKLVC presentation.
[12] Graham, C. 2015. Fundamental Frequency Range in
Japanese and English: The Case of Simultaneous
Bilinguals. Phonetica, 71(4), 271–295.
[13] Grosjean, F. 1992. Another View of Bilingualism. In
Advances in Psychology (Vol. 83, pp. 51–62). Elsevier.
[14] Grosjean, F. 1998. Studying bilinguals:
Methodological and conceptual issues. Bilingualism:
Language and Cognition, 1(2), 131–149.
[15] Keating, P., & Kuo, G. 2012. Comparison of speaking
fundamental frequency in English and Mandarin. The
Journal of the Acoustical Society of America, 132(2),
1050–1060.
[16] Ladd, D. R. 2008. Intonational Phonology (2nd ed.).
Cambridge: Cambridge University Press.
[17] de Leeuw, E. Native speech plasticity in the German-
English late bilingual Stefanie Graf: A longitudinal
study over four decades. Accepted
[18] de Leeuw, E. & Bogulski, C. A. 2016. Frequent L2
language use enhances executive control in bilinguals,
Bilingualism: Language and Cognition, 18(03), 561-
567
[19] Loveday, L. 1981. Pitch, politeness and sexual role:
An exploratory investigation into the pitch correlates of
English and Japanese politeness formulae. Language
and Speech, 24(1), 71–89.
[20] Marian, V., Blumenfeld, H. K., & Kaushanskaya, M.
2007. The Language Experience and Proficiency
Questionnaire (LEAP-Q): assessing language profiles
in bilinguals and multilinguals. Journal of Speech,
Language, and Hearing Research: JSLHR, 50(4), 940–
967
[21] Mennen, I., Schaeffler, F., & Docherty, G. 2008. A
methodological study into the linguistic dimensions of
pitch range differences between German and English.
Proceedings 6th Speech Prosody, Campinas.
[22] Mennen, I., Schaeffler, F., & Docherty, G. 2012.
Cross-language differences in fundamental frequency
range: A comparison of English and German. The
Journal of the Acoustical Society of America, 131(3),
2249–2260.
[23] Müllensiefen, D., Gingras, B., Musil, J., & Stewart, L.
2014. The Musicality of Non-Musicians: An Index for
Assessing Musical Sophistication in the General
Population. PLOS ONE, 9(2), e89642.
[24] Ohara, Y. 1992. Gender-dependent pitch levels: A
comparative study in Japanese and English. In K. Hall,
M. Bucholtz, & B. Moonwomon (eds.), Locating
Power: Proceedings of the Second Berkeley Women
and Language Conference, April 4 and 5, 1992.
[25] Ohara, Y. 1999. Performing gender through voice
pitch: A cross-linguistic analysis of Japanese and
American English. In: U. Pasero & F. Braun (eds.),
Wahrnehmung und Herstellung von Geschlecht:
Perceiving and Performing Gender. Wiesbaden: VS
Verlag für Sozialwissenschaften.
[26] Ohara, Y. 2004 Prosody and Gender in Workplace
Interaction: Exploring Constraints and Resources in the
Use of Japanese, In: S. Okamoto & J.S. Shibamoto
Smith (Eds.) Japanese Language, Gender, and
Ideology: Cultural Models and Real People. New
York: Oxford University Press
[27] Ordin, M. & Mennen, I. 2017. Cross-Linguistic
Differences in Bilinguals' Fundamental Frequency
Ranges. Journal of Speech Language and Hearing
Research, 60 (6), 1493-1506
[28] Passoni, E., Mehrabi, A., Levon, E., de Leeuw, E.
2018. Bilingualism, pitch range and social factors:
preliminary results from Japanese-English sequential
bilinguals. Proceedings 9th Speech Prosody, Poznan
[29] Peirce, J. W. (2007). PsychoPy—Psychophysics
software in Python. Journal of Neuroscience Methods,
162(1–2), 8–13.
[30] R Core Team. 2014. R: A language and environment
for statistical computing. R Foundation for Statistical
Computing. Vienna, Austria Retrieved from
http://www.R-project.org/.
[31] Scherer, K. 2003. Vocal communication of emotion:
A review of research paradigms. Speech
Communication, 40(1–2), 227–256.
[32] Sellars, V. 2000. Anxiety and reading comprehension
in Spanish as a foreign language. Foreign Language
Annals, 33(5), 512-521
[33] Ullakonoja, R. 2007. Comparison of pitch range in
Finnish (L1) and Russian (L2). Proceedins of 16th
ICPhS, Saarbrücken, 1701-1704.