Available via license: CC BY 4.0
Content may be subject to copyright.
CJAL * RCLA Sleeper 52
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
Singing Synthesizers: Musical Language Revitalization through
UTAUloid
Morgan Sleeper
Macalester College
Abstract
Music plays many important roles in language revitalization, from attracting learners and
fostering speech communities to supporting language learning. These effects, however, are
largely independent from the skills which linguists bring to language revitalization. This
study introduces one concrete way in which applied linguistics can directly support musical
language revitalization with UTAUloids – speech-and-music software synthesizers –
illustrated through the creation of a Cherokee UTAUloid as part of ancestral language
reclamation by a learner-linguist Cherokee Nation citizen.
Through their focus on “massive collaboration,” low-resource music production, and youth
involvement, UTAUloids are uniquely situated to serve as instruments for language
revitalization. Even the act of creating an UTAUloid itself allows speakers and learners who
may not consider themselves “musical” to contribute to musical language revitalization, and
this study provides a step-by-step methodology to make creating an UTAUloid as accessible
as possible for anyone interested in incorporating music into their own language
revitalization practice.
Résumé
La musique joue un rôle important pour la revitalisation des langues : attirer des apprenants,
créer des communautés de locuteurs et soutenir l'apprentissage des langues. Or, ces effets ne
font normalement pas partie des compétences que les linguistes apportent à la revitalisation
des langues. Cette étude présente une faon dont la linguistique appliquée peut aider la
revitalisation musicale des langues avec UTAUloids – des synthétiseurs de parole et de
musique – à travers la création d’un UTAUloid Cherokee, un projet de récupération de la
langue ancestrale par un apprenant-linguiste et citoyen de la Cherokee Nation.
Mettant l’accent sur la « collaboration massive, » la production de musique à faibles coûts et
l’implication des jeunes, les UTAUloids sont particulièrement bien placés pour servir
d’instruments de la revitalisation des langues. Même le simple fait de créer un UTAUloid
permet aux locuteurs et aux apprenants qui ne se considèrent pas comme des « musiciens »
de contribuer à la revitalisation musicale des langues. Cette étude propose une méthodologie
pas à pas pour rendre la création d’un UTAUloid aussi accessible que possible à toute
personne souhaitant intégrer la musique dans sa propre pratique de revitalisation des langues.
CJAL * RCLA Sleeper 53
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
Singing Synthesizers: Musical Language Revitalization through UTAUloid
Introduction
Language revitalization is inherently interdisciplinary, and one intersection that has
seen a growing amount of scholarly and community attention in recent years is the
confluence of language revitalization and music. Music can play many roles in language
revitalization, from helping learners progress in terms of grammar, pronunciation, and
vocabulary (Bracknell et al., 2021; Tuttle, 2012; Vallejo, 2019), to reinforcing identity
(Barrett, 2016; Dołowy-Rybińska, 2020; Johnson, 2012; Llewellyn, 2000; Lucas, 2021;
Sparling et al., 2022), expanding a language’s domains of use (Bracknell et al., 2021;
Cotter, 2001; Cru, 2018; Lucas, 2021; Sometimes & Kelly, 2010), bringing speakers and
learners together (Ashton, 2020; Bracknell, 2020; Nummelin, 2020), and infusing language
learning with a powerful sense of joy (Przybylski, 2018; Sparling et al., 2022; Vallejo,
2019). Notably, however, the labour which enables these positive effects of music is
largely independent from the specific skillset linguists bring to language revitalization, and
music has thus far played little role in the productive relationship between applied
linguistics and Indigenous language revitalization (outlined in, e.g., Daniels & Sterzuk,
2022; McIvor, 2020). To that end, this study introduces one concrete way in which these
two disciplines can come together to support musical language revival, through
UTAUloids: free and open-source speech-and-music software synthesizers used for
collaborative vocal songwriting. This study explores their potential for language
revitalization through the creation of an UTAUloid in Cherokee – a revitalizing Iroquoian
language of Oklahoma, the US Southeast, and the Cherokee people dispersed around the
United States and beyond in a digitally connected “at-large” community – as part of
ancestral language reclamation by a learner-linguist Cherokee Nation citizen.
The paper begins with an overview of the Cherokee language and revitalization,
before exploring the place of music in language revitalization efforts more broadly. It then
introduces UTAUloids (along with their inspiration, Vocaloids) before illustrating a step-
by-step methodology for creating an UTAUloid using Cherokee as an example, followed
by a discussion of UTAUloids in language revitalization and the ways in which these
unique musical tools can expand how we conceptualize the relationship between applied
linguistics and Indigenous language revitalization.
Cherokee
Cherokee (ᏣᎳᎩ tsalagi) is an Iroquoian language, part of a family that also
includes Mohawk, Wendat/Wyandot, Seneca, Tuscarora, and Cayuga (Mithun, 1999). As of
the 2010 census, Cherokee counted 12,300 speakers, including approximately 10,000 in
and around the Cherokee Nation in Oklahoma, approximately 1,000 in North Carolina
(where the Eastern Band of Cherokee Indians are located), and an undetermined number of
members of the United Keetoowah Band of Oklahoma and Arkansas (Golla, 2010). The
present-day geographical distribution of Cherokee speakers is a result of a prolonged
campaign of ethnic cleansing by the US government, and particularly of the 1838 forced
removal of the Cherokee people from their homelands in the US Southeast to eastern
Oklahoma, commonly known as the Trail of Tears. The majority of surviving Cherokee
CJAL * RCLA Sleeper 54
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
settled in Oklahoma, in what is now the Cherokee Nation reservation, while groups who
escaped the initial removal and took refuge in the Appalachian Mountains eventually settled
in western North Carolina, the territory of today's Eastern Band of Cherokee Indians.
Beyond these two communities, there is a sizable diasporic population of Cherokee people
across the United States; out of the Cherokee Nation’s 450,000 citizens, around 180,000
reside outside of Oklahoma (ᎠᎾᏗᏍᎪᎢ/Anadisgoi, 2023), with significant populations in
California, Washington State, Texas, Kansas, and Florida.
Typologically, Cherokee is a polysynthetic language and distinguishes itself from
other Iroquoian languages as the only variety in the family with lexical tone. Culturally, the
Cherokee language is strongly associated with its syllabic orthography (Cushman, 2012),
developed by ᏍᏏᏉᏯ (Sequoyah) in the 1820s. The syllabary consists of 85 symbols,
each representing a CV or V syllable of Cherokee; the sole exception is <Ꮝ>, which
represents the single segment /s/. The syllabary is shown with romanized orthographic
equivalents
1
in Table 1:
Table 1
Cherokee Syllabary
Ꭰ 'a'
Ꭱ 'e'
Ꭲ 'i'
Ꭳ 'o'
Ꭴ 'u'
Ꭵ 'v'
Ꭶ 'ga'
Ꭷ 'ka'
Ꭸ 'ge'
Ꭹ 'gi'
Ꭺ 'go'
Ꭻ 'gu'
Ꭼ 'gv'
Ꭽ 'ha'
Ꭾ 'he'
Ꭿ 'hi'
Ꮀ 'ho'
Ꮁ 'hu'
Ꮂ 'hv'
Ꮃ 'la'
Ꮄ 'le'
Ꮅ 'li'
Ꮆ 'lo'
Ꮇ 'lu'
Ꮈ 'lv'
Ꮉ 'ma'
Ꮊ 'me'
Ꮋ 'mi'
Ꮌ 'mo'
Ꮍ 'mu'
Ꮎ 'na'
Ꮏ 'hna'
Ꮐ 'nah'
Ꮑ 'ne'
Ꮒ 'ni'
Ꮓ 'no'
Ꮔ 'nu'
Ꮕ 'nv'
Ꮖ 'qua'
Ꮚ 'que'
Ꮘ 'qui'
Ꮙ 'quo'
Ꮚ 'quu'
Ꮛ 'quv'
Ꮝ 's'
Ꮜ 'sa'
Ꮞ 'se'
Ꮟ 'si'
Ꮠ 'so'
Ꮡ 'su'
Ꮢ 'sv'
Ꮣ 'da'
Ꮤ 'ta'
Ꮥ 'de'
Ꮦ
'te'
Ꮧ 'di'
Ꮨ 'ti'
Ꮩ 'do'
Ꮪ 'du'
Ꮫ 'dv'
Ꮬ 'dla'
Ꮭ 'tla'
Ꮮ 'tle'
Ꮯ 'tli'
Ꮰ 'tlo'
Ꮱ 'tlu'
Ꮲ 'tlv'
Ꮳ 'tsa'
Ꮴ 'tse'
Ꮵ 'tsi'
Ꮶ 'tso'
Ꮷ 'tsu'
Ꮸ 'tsv'
Ꮹ 'wa'
Ꮺ 'we'
Ꮻ 'wi'
Ꮼ 'wo'
Ꮽ 'wu'
Ꮾ 'wv'
Ꮿ 'ya'
Ᏸ 'ye'
Ᏹ 'yi'
Ᏺ 'yo'
Ᏻ 'yu'
Ᏼ 'yv'
The syllabary was quickly adopted and remains widely used among speakers and learners
today; it has also been incorporated into Unicode since 1999 and is available as a default
input method on Apple's macOS and iOS and Microsoft's Windows operating systems.
In terms of vitality, Cherokee is considered “definitely endangered” by UNESCO
(Moseley, 2010), and though usage by speakers under 40 is low, with most children no
longer learning it as a home language (Cherokee Nation, 2003, as cited in Uchihara, 2016)
the Cherokee people and the Cherokee Nation are committed to revitalizing the language.
Revitalization efforts for Cherokee are multifaceted and wide-reaching, including the
CJAL * RCLA Sleeper 55
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
popular ᏣᎳᎩ ᏧᎾᏕᎶᏆᏍᏗ (Tsalagi Tsunadeloquasdi) immersion schooling from pre-
school through sixth grade (Peter et al., 2017), master/apprentice programs, and community
and university language classes. Media also plays an important role, with radio broadcasts
from the Cherokee Nation and the monthly bilingual Cherokee Phoenix newspaper
providing news and features in the language.
The Cherokee Nation has in recent years also been at the forefront of innovative,
Indigenous approaches which conceptualize language revitalization expansively, in terms of
the lifeways of current and future speakers. The Speaker Services program launched in
2022, for instance, aims to help Cherokee speakers with everyday basic needs, including
home repairs, accessing healthcare and medical devices, and installing new appliances and
accessibility aids in speakers’ homes. This program is housed in the Cherokee Nation’s
Language Department on the belief that “if speakers aren’t worried about their roofs
leaking, they can worry about their grandchildren speaking Cherokee” (Cherokee Phoenix
Staff, 2022). Other recent expansive language revitalization efforts have included installing
a state-of-the-art cellular tower in Kenwood – a remote reservation community of around
1,000 people with a high proportion of fluent, daily speakers – so that young people can
take advantage of remote work to stay in the community, Cherokee speakers can more
readily use the language with friends and family across the Nation, and online language
classes can be taught from the town (Caldwell, 2023).
Another notable aspect of Cherokee language revitalization in the Cherokee Nation
is that there has been a focus on reaching heritage speakers and tribal members who are
physically removed from the reservation and its speech community. The Cherokee Nation
offers several online language classes (at three skill levels) taught by a native speaker
throughout the year, for instance, as well as self-paced pre-recorded classes, and the option
for remote learners to call in to talk with native speakers. A monthly livestream series called
“ᏣᎳᎩ: Wherever we are” was also started in 2021 to connect diasporic Cherokee Nation
citizens with cultural programming on topics like Cherokee spirituality, music, games, and
history, alongside updates on important current tribal issues, including language
revitalization. In addition, the Cherokee Nation organizes annual in-person outreach events
to more than 20 “At-Large” communities with significant populations of tribal members
across the United States, from the Puget Sound and California to Texas, Kansas, and
Florida. Along with voter registration, healthcare information, games, storytelling, and
dance, these annual events also prominently include Cherokee language resources,
materials, and taster classes intended to garner interest and participation in language
preservation from at-large members.
It is within this wider at-large community that I situate both myself and my
relationship to the Cherokee language. A citizen of the Cherokee Nation, I grew up in
Florida, removed from the reservation but continually hearing the importance of our
language, stories, and culture from my father and grandfather, spending time in the
Cherokee Nation while visiting family over summers, and later attending At-Large
gatherings when living in California and Washington State. Neither my father nor my
grandfather spoke Cherokee fluently, but they passed down individual words and phrases –
ᎣᏏᏲ, ᏩᏙ, ᎣᏍᏓ, ᏙᎾᏓᎪᎲᎢ – and a reverence for the language that inspired me to try to
learn, and to pass on words to my son in turn. I have been slowly working towards
reclaiming Cherokee over the past 10 years, through online classes and At-Large meetups.
Now living in Minnesota, the At-Large gatherings are less accessible, and I find myself
CJAL * RCLA Sleeper 56
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
relying on digital means to stay connected to the language and the Cherokee Nation,
including the “ᏣᎳᎩ: Wherever we are” programming, social media, and online language
courses. I am also a lifelong musician – a fact which guides my work as a linguist, and
shapes how I interact with language learning and reclamation; I love to learn through song,
and I have been particularly inspired by other Native people learning their languages
through songwriting (e.g., Przybylski, 2018; Tuttle & Lundström, 2015). This current
project sits at the confluence of these aspects of my identity, and of my relationships with
music, linguistics, and the Cherokee language – UTAUloid gives me a way to use music in
my personal language reclamation, while also leveraging my linguistic skills to help with
wider Cherokee language revitalization efforts, within the context of our digitally-
connected diasporic Cherokee community.
Music in Language Revitalization
This section will provide a brief overview of some of the many ways music can aid
in language revitalization efforts, focusing on three specific areas: attracting learners and
retaining speakers, creating opportunities for speech community, and supporting language
learning.
Attracting learners and engaging interest in the language movement among speakers
is a particularly important contribution music can make in language revitalization. The
Cherokee Nation's Cherokee National Youth Choir, for instance, sings an all-Cherokee-
language repertoire specifically to both introduce young learners to the language, and to
keep immersion school students engaged with the language movement outside of classes
(Cherokee Nation, 2022), while the Oklahoma Native American Youth Language Fair
(ONAYLF) offers a venue for young learners and speakers to share both traditional and
newly composed music in Native languages. For Cornish – a revitalizing language spoken
in Cornwall – music is routinely cited as one of the main ways new learners first encounter
the language, from a folk music revival in the 1970s (MacKinnon, 2005) to the modern
electronic-pop of Gwenno Saunders, who the Cornish Language Board directly credits for
the record number of learners registering for Cornish exams after her 2018 album Le Kov
(BBC News, 2018). Music can also help spread awareness of both revitalizing languages
and Indigenous language revitalization in general to the wider public, as when a translated
cover of The Beatles’ Blackbird in Mi’kmaq by Cape Breton teenager Emma Stevens went
viral in 2019. The video was uploaded to mark the United Nation’s International Year of
Indigenous Languages and has seen over 1.8 million views on YouTube, introducing
viewers to the UN’s campaign for language revitalization while also “[showing] non-
Mi’kmaw people the beauty of our language” (Goodyear, 2019).
Another important function of music in language revitalization is that it can serve to
change people’s attitudes about language, with important benefits for language use. In
Aotearoa New Zealand, for instance, the 1982 song Poi E – a fusion of Māori culture and
hip-hop music co-written in Te Reo Māori by linguist Ngoingoi Pēwhairangi and Maui
Dalvanius Prime – was conceptualized as a way to inspire young Māori listeners to use Te
Reo in their everyday lives (Archer, 2002). The song was a breakout hit and instrumental
part of the Māori language and cultural revival that followed, and helped shape the image
of Māori as a modern, vitally relevant language for a new generation (Sheehan, 2016, p.
78). Elsewhere, in Minnesota, Anishinaabe artist Tall Paul similarly uses bilingual hip-hop
CJAL * RCLA Sleeper 57
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
music to change attitudes about Anishinaabemowin and inspire more young people to take
up the language, saying specifically: “If I incorporate the language into hip hop, it’ll make
the language cool for those kids. Maybe they’ll be interested in learning it at a young age”
(Przybylski, 2018, p. 388).
Music can also be instrumental in language revitalization by creating opportunities
for speech community. In many minoritized language contexts, a prevailing issue is that
another language (or set of languages) has become the medium of everyday
communication; even with a sizable number of fluent speakers, this can make it difficult for
speakers to find natural domains of use for the language in daily life (Hinton, 2001). Music
can help provide this space – both physical and conceptual – for speakers to come together
in speech community.
MacKinnon (2005, p. 249), for instance, notes how Cornish music festivals are
particularly important as “opportunities for Cornish speakers and learners to come together
and use the language”, and concerts, festivals, and participatory musical gatherings fulfill
this purposes in revitalizing language communities around the globe, from the Eisteddfodau
festivals in Wales to Breton Fest-Noz (“night-festivals”) (Dołowy-Rybińska, 2020), the
multi-Nation sākihiwē indigenous music festival in Winnipeg (Przybylski, 2021),
Guernésiais choir rehearsals on Guernsey (Johnson, 2012), and the concerts of the Ainu
band Marewrew. Marewrew perform entirely in Ainu – a critically endangered language
indigenous to Hokkaido – and explicitly teach the language behind their repertoire as part
of their performances, explaining translations and linguistic concepts while teaching the
audience to sing along. This approach creates a space where “the audience become active
participants in a performance that uses the Ainu language… almost delivered as mini-
workshops” (Nummelin, 2020, p. 291).
In addition to physical gatherings of speech community, music can also help to link
speech community through distributed media, such as radio and podcasts. Radyo an
Gernewegva (literally “radio of the Cornish-language area”), for instance, is a Cornish-
language, podcast-format radio show produced weekly since 2007, with the goals of
providing Cornish immersion, showcasing Cornish-language music, culture, and talent, and
giving Cornish musicians exposure. Its listeners are often language learners, and an active,
online community forum encourages participation and engagement with the music in
Cornish, even when participants are geographically isolated from other speakers.
Traditional over-the-air radio in revitalizing languages fulfills a similar function, and is
particularly effective at “strengthening, sustaining, and revitalizing cultural and linguistic
traditions” (Danos & Turin, 2021, p. 76), by creating asynchronous “speech communities”
that can be tuned into even as listeners go about their daily lives.
Finally, perhaps one of the most immediately obvious ways in which music can
contribute to language revitalization is in supporting language learning. Of course, this
benefit is not restricted to revitalizing languages, and much of the research in this area is
based on commonly taught languages (Davis, 2017; Engh, 2013; Good et al., 2015; Tegge,
2018); but the central place of language teaching and learning in revitalization makes this
affordance of music especially relevant here.
A large portion of previous research on music in language teaching focuses on the
use of songs in classroom pedagogy and describes how music can be particularly helpful
for teaching natural pronunciation, new grammatical structures, and idioms (Jolly, 1975);
capturing students’ attention and contextualizing colloquial uses of language (Abrate,
1983); and introducing vocabulary domains and language issues not commonly brought up
CJAL * RCLA Sleeper 58
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
in textbooks (Schmidt, 2003). More recently, Vallejo (2019) explores how teachers in
Kanien’ke:ha (Mohawk) language immersion programs use both traditional and
contemporary songs to provide wholistic, culturally-grounded language education, with
music as a “linchpin pedagogical tool that promotes intergenerational interactions, builds
social relationships, and facilitates the daily use of language in and outside the classroom.”
Notably, the musical repertoire used includes translations of Western nursery rhymes,
popular hits from Elvis and Johnny Cash, and Christmas carols alongside traditional
Kanien’ke:ha music, and while these may on the surface seem incompatible with the idea
of a culturally-grounded Kanien’ke:ha language education, Vallejo (2019) argues that
English-language songs essentially “take on a new form” when translated into
Kanien’ke:ha (p. 106), and that this mix is both pedagogically beneficial and reflective of
the broad musical tastes and experiences of the Kanien’ke:ha community.
Of course, many revitalizing languages are not taught in classroom settings, and
music can help support language learning in these situations as well, especially in terms of
indigenous methodologies. Antoine (2015) shows how Lakota songs can teach language
and culture “from the Native perspective, reinforce culturally appropriate ways of behavior,
and teach the tribe's social structure as well as history and spirituality” (p. 17). Tuttle and
Lundström (2015) present a case study of a young composer learning potlatch singing in
three endangered Interior Athabaskan languages of Alaska directly from elders and point
specifically to how “gaining proficiency in song, through Indigenous channels, can further
proficiency in language” (p. 38)
One important thread running through many of these musical approaches is that
they are not exclusively – or even primarily – concerned with a narrow view of language
revitalization as “increasing speaker numbers.” And in fact, many of these musical practices
might be more accurately thought of as forms of language reclamation, defined by linguist
Wesley Leonard as “a larger effort by a community to claim its right to speak a language
and set associated goals in response to community needs and perspectives” (Leonard, 2012,
p. 359). Rather than revitalizing a “language” in the abstract, language reclamation is a
process of “personal and communal agency and the expression of Indigenous identities,
belonging, and responsibility to self and community” (McCarty et al., 2018, p. 160), which
includes claiming and creating new practices around language and music and (re-)defining
culture (Leonard, 2012).
Each of the three areas above, then – attracting and engaging learners and speakers,
fostering speech communities, and supporting language learning – are important ways in
which language revitalization and reclamation can happen through music. Notably,
however, the necessary work which enables this – composing, producing, and performing
music, organizing concerts and festivals, documenting song repertoires – is not something
that linguists are trained to help with. To that end, the remainder of this study introduces a
practical overlap between applied linguistics and Indigenous language revitalization in
which these two fields can collaboratively create a powerful tool for musical language
revitalization: vocal music production tools known as UTAUloids.
Vocaloid
To fully explain UTAUloids, it is first necessary to briefly introduce their
inspiration, Vocaloids. Often described as “anthropomorphized singing synthesizers,”
Vocaloids were developed by Yamaha and the Music Technology Group at Universitari
CJAL * RCLA Sleeper 59
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
Pompeu Fabra, Barcelona in 2004 (Kenmochi & Ohshita, 2007). Vocaloid software uses a
library of human speech samples (called a “voicebank”) and combines them with musical
information in order to output digitally “sung” melodies. Voicebanks are created by
recording a voice donor (usually a singer or voice actor/actress) singing all possible
syllables (or morae) in a language at several different pitches, so that each individual
sample can be arranged and manipulated by the Vocaloid synthesizer to approximate a
singing voice.
2
The first Vocaloids were developed to sing in English and Japanese, but
subsequent releases have expanded to include languages like Mandarin, Korean, and
Spanish.
Users of the Vocaloid software input both notes and syllables using a piano roll-
style score editing interface shown in Figure 1, which features an excerpt from the Japanese
folk song Sakura, Sakura:
Figure 1
Vocaloid Score Editor (Clusternote, 2014)
The Vocaloid software then takes the resulting score as input into its synthesis engine,
which chooses appropriate samples from its voicebank (i.e. the syllables to be sung, at the
nearest pitch to the target note) and concatenates and alters these samples based on musical
information in the score (including pitch, dynamics, timbre, attack, decay, reverb, and
vibrato) to produce the synthesized output: a sung melody. This synthesis engine is
represented graphically in Figure 2, adapted from Kenmochi and Ohshita (2007).
This is the basic synthesis behind the sounds the Vocaloid software produces, but an
equally important component of the Vocaloid concept is that the individual packages are
anthropomorphized; they are not just 'singing synthesizers', but “synthesized singers.”
There have been dozens of different Vocaloids introduced since the technology's debut in
2004, but by far the most popular has been Hatsune Miku, designed by Crypton Future
Media in Japan.
Hatsune Miku (初音ミク, meaning “first sound of the future”) made her debut in
2007, and is stylized as a “virtual idol,” with long turquoise twintails and design features on
her costume which recall the user interface of her software.
CJAL * RCLA Sleeper 60
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
Figure 2
Vocaloid System Diagram (adapted from Kenmochi & Ohshita, 2007)
She has proved enormously successful; over 100,000 songs have been created using her
voicebank, and she appears in over 170,000 YouTube videos, dozens of which have more
than a million views each
3
(Crypton Future Media, n.d.). She has starred in video games
and car commercials, topped album charts in Japan, and performed (via hologram) as an
opener for Lady Gaga and in her own headlining sold-out concerts around the world. Her
rise to become something of an international icon illustrates how the creative culture which
has arisen around Vocaloids makes them much more than just instruments for the
producers, musicians, and fans who use them.
One of the key pieces of Vocaloid's appeal is the fact that Vocaloid encourages
iterative creation by multiple users over the internet. The specific combination of format
(exportable so that users can download and edit notes, lyrics, and other parameters on
existing tracks) and licensing (which in general allows for the free personal and commercial
use of Vocaloids’ voices and likenesses in videos, art, and other adaptations) has made
Vocaloid especially conducive to what has been called “massive collaboration” (Sousa,
2014).
This “massive collaboration” means that users can interact with and iterate on
Vocaloid content in a myriad of ways. One person might upload a melody composed in the
software; a second person could then download it, add drum and bass tracks behind it, and
re-upload it. A third person could remix the music, change the lyrics, and then re-upload the
result. A fourth person could record themselves singing the tune over the instrumental
backing tracks, and a fifth person could create a video compiling different interpretations of
the song.
This phenomenon is a central component of Vocaloid culture, and one result is that
the boundary between “fan” and “creator” in Vocaloid is substantially blurred, if not
altogether erased. Rather than an audience passively consuming musical texts, Vocaloid
users are “a distributed group of fan-producers” (Condry, 2011), and Vocaloid has become
CJAL * RCLA Sleeper 61
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
“a catalyst for collective, grassroots, and multidisciplinary creation” (Sousa, 2014)
spanning music, lyrics, art, animation, costume design, choreography, writing, and more.
The potential for 'massive collaboration' is a large part of Vocaloid's appeal to its
users and fans, along with the inherent creative control it provides; Vocaloid producers can
fine-tune almost every part of a performance, from backdrops and costumes to vocal
delivery, intonation, and microtiming. The only limitation to what can be customized in a
Vocaloid performance comes from the voicebank itself – the voice donor's initial
recordings. This is where UTAUloid comes in.
UTAUloid
UTAUloid
4
(From 歌う utau meaning “to sing” in Japanese), created by developer
Ameya/Ayame in 2008, is a freeware implementation of the Vocaloid concept. Like
Vocaloid, UTAUloid allows users to key in scores with notes and lyrics, which it in turn
relays to a synthesis engine to produce synthesized singing output. The main difference
between Vocaloid and UTAUloid is that while Vocaloid products come with a pre-recorded,
non-modifiable voicebank, UTAUloid instead allows users to record and use their own
voicebanks, which then act as a sample library for the synthesis. Importantly, this means
that not only can UTAUloids sing in any voice, they can also sing in any language.
In addition to UTAUloids who sing in Japanese and English, users have created
UTAUloids singing in many languages not available in any Vocaloid, including Catalan,
Tagalog, Indonesian, Basque, Irish, and Esperanto. This is particularly impressive because
creating an UTAUloid requires a significant amount of specialized linguistic knowledge
about a given language, and this is reflected in the linguistic proficiency of the wider
community. Fans on the UTAU Wikia, UTAU Wiki, and UtaForum.net websites create
linguistic tutorials, curate threads on relevant phonetic and phonological issues for specific
languages, recommend scholarly linguistic work as references, and synthesize academic
articles into advice for creators to use when recording their “reclists” of potential phonemic
combinations.
Along with this community linguistic knowledge, it is notable that the linguistic
plurality of UTAUloids available includes several who sing in minoritized, marginalized,
and revitalizing languages. One prominent example is Sachi Eika (詠歌サチ), an Irish-
language UTAUloid created by user Jadii in 2009. Figure 3 shows Sachi Eika's design, in
both the key art and a 3D model for use in music videos.
As with Vocaloids, UTAUloids like Sachi Eika exist within the framework of
massive collaboration and collective creation (Le, 2014); once made available online,
UTAUloid fans were able to write and produce music with Sachi Eika, re-record existing
Vocaloid and UTAUloid songs with her voice, download and modify others' Sachi Eika
songs, and create art, videos, and other examples of “Nth fanfiction” (Kenmochi, 2010)
featuring the character. As of 2023, YouTube hosts over 140 videos tagged for Sachi Eika;
the fan-art community website DeviantArt listed over 100 works of the character before its
recent restructuring, and similar Japanese site Pixiv still hosts several dozen; Sachi Eika
also appeared as a character in a printed Irish manga magazine in 2010 (UTAU Wikia
contributors, n.d.).
CJAL * RCLA Sleeper 62
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
Figure 3
Sachi Eika (Jadii, 2009)
Additionally, while most of these fan works were created in the years directly
following Sachi Eika's debut, the nature of UTAUloid means that even over a decade later,
it is still possible to load Sachi Eika's voicebank onto any computer and have her sing,
within a matter of minutes, an Irish-language song, such as Báidín Fheilimí, shown in
Figure 4 below.
Figure 4
Sachi Eika UTAUloid Score for Báidín Fheilimí
CJAL * RCLA Sleeper 63
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
It is in this vein that UTAUloids can be used as literal and figurative instruments for
language revitalization, and the remainder of this article provides a step-by-step guide for
producing an UTAUloid for revitalization, illustrated through the creation of an example
Cherokee UTAUloid called ᎧᏃᎩᏍᏗ (kanogisdi “singing”).
Method: Building a Cherokee UTAUloid
This section will demonstrate the procedure for creating an UTAUloid, illustrating
the process through Cherokee. An UTAUloid can be broadly conceptualized into three
parts: the synthesis engine, the sample library (“voicebank”), and the set of instructions
(called “tunings”) unique to each UTAUloid that specify how the synthesis engine should
interpret, blend, and concatenate individual samples into synthesized singing. The three
main components of creating an UTAUloid, then, are to install the synthesis engine, to
record a voicebank of samples, and to tune those samples for synthesis; and these final two
steps can be directly conceptualized as applied linguistic work.
The UTAU synthesis engine is available in several different software packages,
including the original UTAU application for Windows,
5
UTAU-Synth for macOS,
6
and
more recently, the open-source project OpenUTAU,
7
compatible with both Windows and
macOS. Voicebanks and files are fully cross-compatible between the different applications,
and the user interfaces are also largely the same. The Cherokee UTAUloid and examples
shown here were created in OpenUTAU, which is recommended for several reasons: it is in
active development by an engaged, helpful, and welcoming community; its interface is
fully localized into 17 languages; and its open-source license and design allows for all the
benefits of using open tools in Indigenous language revitalization, including accessibility,
longevity, and community customization (Berez-Kroeker et al., 2023; Brinklow et al., 2019;
Salazar et al., 2021).
Recording a Voicebank
Recording a voicebank of samples from the UTAUloid's voice donor is a process
that requires the application of a certain amount of linguistic knowledge about the target
UTAUloid language, and closely resembles collecting wordlist data for phonetics research.
The first step is to assemble the list of sounds to be recorded, into what the
UTAUloid community calls a “reclist.” The reclist should consist of all possible
combinations of phonemes needed to reproduce any syllable (or mora) in the target
language. This requires a linguistic understanding of the phonetic inventory, phonotactics,
and syllabic (or moraic) structure of the language, as well as an understanding of how
UTAUloid synthesis works in order to know exactly what phoneme combinations need to
be recorded.
The phonetic inventory of Cherokee (Montgomery-Anderson, 2015; Pulte &
Feeling, 1975; Uchihara, 2016) is presented in Tables 2 and 3 below, based on
Montgomery-Anderson (2015) and Uchihara (2016):
CJAL * RCLA Sleeper 64
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
Table 2
Cherokee Consonant Inventory
BILABIAL
ALVEOLAR
PALATAL
VELAR
LABIO-
VELAR
GLOTTAL
PLOSIVES
t
k
kw
ʔ
AFFRICATES
CENTRAL
ʧ
LATERAL
tɬ
FRICATIVES
s
h
NASALS
m
n
APPROXIMANTS
CENTRAL
j
w
LATERAL
l
Table 3
Cherokee Vowel Inventory
FRONT
CENTRAL
BACK
HIGH
i
iː
u
uː
MID
e
eː
ə
əː
o
oː
LOW
a
aː
From this inventory, the next step is to determine how the phonemes pattern in terms of
syllables (or moras), which serve as the unit of recording for UTAUloid. In Cherokee, the
minimal syllable consists of a vowel, with optional onsets of single consonants or more
complex clusters, and an optional single consonant coda, represented in Table 4 (adapted
from Uchihara, 2016) below:
Table 4
Cherokee Syllable Structure
ONSET
NUCLEUS
CODA
(s)
(t ʧ tɬ l k kw)
(h)
V(V)
(l n j w ʔ h)
(n j w)
(s m h ʔ)
Even with Cherokee's relatively small phonetic inventory, this maximal syllable
structure would make recording each possible potential syllable for sampling a considerable
task. Thankfully, the architecture of UTAUloid allows for combining samples on a single
note, so that a smaller number of recorded samples can approximate the entire syllabic
inventory of a language. CVC syllables, for instance, can be produced by combining two
appropriate CV and VC segments, and in languages with phonemic vowel length (such as
Cherokee), it is not necessary to record separate samples for V and V: (or CV and CV:)
syllables; vowels can be lengthened either by note length or with the addition of a
CJAL * RCLA Sleeper 65
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
following sample of the same vowel quality. Likewise, complex onsets and codas can be
'cropped' via volume manipulation in the UTAU score, so that a CaCbV segment could also
be used to render a CbV syllable, and a CxV syllable and CyV syllable could be manipulated
to form the complex onset of a CxCyV syllable.
Of course, because of the phonetic effects of co-articulation and gestural timing, the
more an UTAUloid voicebank relies on 'shortcuts' such as syllable cropping, the less natural
it will sound; in general, the more unique segments are recorded, the more ‘natural’ the
UTAUloid will sound. It is worth noting, however, that “natural” is not necessarily the goal
of UTAUloid, and many popular UTAUloids and Vocaloids are made without taking every
phonetic process of a language into account. Hatsune Miku, for instance, does not produce
the vowel devoicing between voiceless consonants which is saliently characteristic of much
spoken Japanese, and deviation from “natural” phonetic production is often seen as a
positive creative affordance of UTAUloids and Vocaloids.
Another consideration is that, particularly in Indigenous language revitalization
contexts, it may be important to record samples that are not strictly necessary for forming
words from the language’s phonemic inventory but are important for other reasons. In
Cherokee, for instance, the syllables /hna/ and /nah/ would not be part of a minimal reclist,
since they can be formed by combining other CV and VC samples (i.e. /na/ and /ah/). Both
of these syllables are, however, also independent characters in the Cherokee syllabary (Ꮏ
'hna' and Ꮐ 'nah'), and because of the cultural importance of the syllabary, it was a goal of
this UTAU project to include each character as a standalone sample, and so they were
recorded here as well.
Taking the above factors into account, the reclist for the Cherokee UTAUloid
consists of the following segments:
Table 5
Segments for Cherokee Reclist, by Syllable Type
Category
Examples
Number
1
CV: all initial consonants + all vowels
/tɬo/ /se/ /ʔa/
78
2
CCV: /s/ + {/k kw t ʧ tɬ l/} + all vowels
/skwa/ /stə/ /stɬi/
36
3
CCV: {/k kw t ʧ tɬ l n j w/} + /h/ + all
vowels
/khi/ /tho/ /nhe/
54
4
V: all vowels
/a/ /e/ /i/
6
5
VC: all vowels + final {/l n j w ʔ h/}
/əl/ /aʔ / /uh/
36
6
Misc.: Ꮏ 'hna' and Ꮐ 'nah'
/hna/ /nah/
2
Total:
212
Once the reclist is complete, it can then be recorded. The choice of voice to be
recorded is an important one, and will have the most significant impact on the sound of the
finished UTAUloid. In many language revitalization contexts, older speakers and/or skilled
language users (as recognized within the community) may have the phonetic productions
considered most representative or most desirable by speakers and learners. On the other
hand, the act of recording for an UTAUloid voicebank is also a uniquely approachable
CJAL * RCLA Sleeper 66
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
language project for heritage learners, who do not need to be even close to fluent or
conversational in their language to create a fully-featured UTAUloid; all that is required is a
willingness to produce the required sounds. Recording a voicebank can be a powerful act of
language reclamation for heritage learners, then, both in terms of the practical
pronunciation practice it entails, and in the ability to contribute to language revitalization in
a significant way even early on in the language learning process. Ultimately, any speaker or
learner of a language can voice an UTAUloid,
8
and recording multiple UTAUloids in a
language is, of course, both feasible and beneficial. For this Cherokee UTAUloid example,
I provided the voice recordings, as part of my own language reclamation process as a
heritage learner of Cherokee.
For recording, UTAU requires samples to be in WAV format, but these can come
from any source. They can be recorded directly onto a computer using free software like
Praat (Boersma & Weenink, 2023) or Audacity (Audacity Team, 2022), or on a dedicated
recording device and copied to the computer for use in UTAU. For this Cherokee
UTAUloid, the reclist was recorded as a single WAV file in Praat, at 44.1 kHz in 16bit,
using an Audio-Technica ATR2100-USB dynamic cardioid microphone, and then split into
individual samples using Audacity.
Aside from a quiet environment and the use of a windscreen if feasible – to cut
down on high frequencies which can make it difficult for the UTAU synthesis engine to
blend sibilants smoothly – the main considerations when recording specifically for
UTAUloid are that each syllable or mora should be as close to a steady pitch as is possible
for the speaker (i.e. trying to avoid list intonation), and that the nucleus of the syllable or
mora should be held long enough to ensure that the synthesis engine has enough steady
state of the nucleus to adjust as needed. Figure 5 illustrates raw samples for the sequences
/ta kʷa ʧa/ of the Cherokee UTAUloid:
Figure 5
Raw Samples for /ta kʷa ʧa/
After all raw audio files are recorded, they can then be 'tuned' for synthesis.
CJAL * RCLA Sleeper 67
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
Tuning Samples for Synthesis
Once the raw audio has been recorded, the next step in turning these samples into a
viable voicebank is to “tune” them for use in UTAU, by demarcating specific regions of
time in each individual sound file that are relevant for manipulation by the synthesis engine
– a process which directly applies linguistic skills in phonetics and spectrogram reading.
The first step to tuning is to create a plain-text file called “oto.ini”
9
in the voicebank
directory listing each individual .wav file on its own line, followed by an “=,” after which
the tunings can be stored. For example, an excerpt from the Cherokee UTAUloid’s blank
“oto.ini” file before tuning would read:
o.wav=
go.wav=
yo.wav=
The tuning process then varies slightly by the UTAU software used,
10
but in OpenUTAU, it
can be done through the voicebank editor in the “Singers” tool. The voicebank editor,
shown in Figure 6, lists each individual .wav file present in the voicebank and displays a
waveform and spectrogram of the selected sound, along with a series of editable attributes
used in tuning: [1] alias, [2] offset, [3] consonant,
11
[4] cutoff,
12
[5] pre-utterance, and [6]
overlap.
Figure 6
Voicebank Editor in OpenUTAU
CJAL * RCLA Sleeper 68
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
Alias ([1]) allows users to specify one or more alternate names for a given sound. This can
be used to allow for samples to be referenced using either phonetic transcriptions or
orthography, or for the use of more than one orthographic system in a single voicebank. In
Cherokee, for instance, adding an alias could allow for a sample “o.wav” (/o/) to be called
with either <o> or <Ꭳ> (the syllabary character for /o/) in the score.
The rest of these attributes variously come into play while tuning, as the process
varies slightly depending on the structure of the syllable or mora being tuned, with one
method for syllables consisting only of vowels, and another for any syllable or mora
containing any number of consonants. There are also different philosophies, personal
preferences, and aesthetic considerations that can affect how a creator might tune their
UTAUloid, and so rather than an exhaustive survey of UTAU tuning in general, the
following subsections detail the particular tuning process used in this Cherokee UTAUloid,
beginning first with vowels, and then syllables containing consonants.
Vowels
Figure 7 shows an example of tuning of a syllable consisting of only vowels
(whether V, V:, VV, etc.), illustrated with the Cherokee syllable Ꭳ /o/:
Figure 7
Tuning a V Syllable (
Ꭳ
/o/)
CJAL * RCLA Sleeper 69
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
For vowels like Ꭳ /o/, the relevant attributes which can be edited in tuning are the offset
(indicated by the blue-shaded region on the left [1]), the steady state (the central white
region [2]), the cutoff (the blue-shaded region on the right [3]), and the overlap (indicated
by the line in [4]). The offset [1] is measured in time in milliseconds relative to the start of
the file and indicates where the sample will begin to be played from in synthesis in order to
omit any preceding silence and often the initial, volatile state of the vowel. For vowels, the
offset can be positioned at the beginning of a periodic cycle in the steady state, as seen in
Figure 7 above.
The next modifiable attribute is the cutoff [3], measured in milliseconds from the
end of the file, which indicates where the playback will end in synthesis, and is used to
exclude the end or decay of the sound, as well as any following silence. For vowels, this
can be placed at the end of a periodic cycle in the steady state, with the result that the
steady state [2] – the region of the sample the synthesis engine will stretch or shrink to fit
the required note length – should ideally represent a loopable, periodic series of cycles.
The final attribute for tuning vowels is the overlap [4]. Measured in milliseconds
relative to the offset, it indicates how far the synthesis engine should cross-fade the
previous note into a sample; the portion of the sample to the left of the overlap will be
mixed with the previous sample, while the portion to the right will not. There is no specific
point of the vowel that the overlap needs to be anchored to, but 50ms after the offset results
in a blending sound consistent with popular UTAUloid voicebanks.
Consonants
For segments containing any number of consonants (whether C0V, VC0, etc.), the
same attributes apply, with the additional considerations of a “consonant” region and a pre-
utterance point. Figure 8 shows the tuned Cherokee syllable Ꭺ /ko/.
The offset [1] performs the same function here as in the V example above: marking
where the sample will begin to play, and excluding the region shaded in blue from
synthesis. The cutoff [3] again does the same for marking the end of the sample, by
excluding the blue region, and should extend to the end of a periodic cycle of the vowel, so
that the steady state [2] represents a loopable periodic vowel sound.
The difference for syllables with consonants is that instead of being bound by the
offset [1] and cutoff [3] points, the steady state [2] is here bound by the cutoff [3] and a
“consonant” region [5]. The length of the consonant region [5] is measured in milliseconds
relative to the offset, and this indicates the portion of the sound file that will be played in
synthesis, but not manipulated in terms of length by default
13
. The entire consonant region
will be played, without being either stretched or shrunk to account for note length as the
steady state [2] is. In tuning consonant syllables, then, this region should include both the
consonant itself and also the initial portion of the vowel affected by formant transitions –
both perceptual cues which are most salient when unaltered by the synthesis engine. For the
same reason, the overlap [4], which determines how far into a sample any cross-fade with a
previous sample should extend to, should be placed before the consonant information, so
that relevant consonant cues (such as the stop burst in Ꭺ /ko/ above) are not obscured or
lost in blending with the previous sample.
CJAL * RCLA Sleeper 70
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
Figure 8
Tuning a CV Syllable (
Ꭺ
/ko/)
The other new attribute for consonant syllables is the red pre-utterance line [6].
Measured in milliseconds relative to the offset, the pre-utterance indicates which point in
the sample should be aligned with the beginning of the sung musical note in terms of
rhythm. This is important because when speakers sing, rhythmic timing is organized around
the nucleus of the syllable (or mora), rather than the onset; the pre-utterance allows users to
align the start of the nucleus with the start of the note. In a CV syllable like Ꭺ /ko/ above,
then, the pre-utterance should be positioned at the onset of the vowel. In some cases, such
as when the preceding consonant is a liquid or glide, it can be difficult to determine the
exact onset of a vowel from the waveform alone. To tune these samples more accurately, it
is useful to refer to the spectrograms shown below the waveform, as seen in Figure 9 which
shows the Cherokee syllable Ᏺ /jo/:
After adjusting each of these parameters for a given sample, clicking “Save Otos”
saves the tunings to the “oto.ini” file in the voicebank directory. This file will then contain
the tuning information for all the sound files in the voicebank, in the format of
[filename]=[alias],[offset],[consonant],[cutoff],[pre-utterance],[overlap], with values in
milliseconds.
CJAL * RCLA Sleeper 71
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
Figure 9
Tuning
Ᏺ
/jo/
The lines of the Cherokee UTAUloid oto.ini corresponding to the examples shown
above after tuning, for example, are:
o.wav=o,224.9,0.0,-566.3,0.0,50.0
go.wav=go,517.0,85.3,-502.5,45.2,17.7
yo.wav=yo,454.4,193.2,-601.4,107.8,47.3
Since the “oto.ini” file is plain text and user-editable, this also means it is possible
to tune the samples for an UTAUloid without using the OpenUTAU software, by
determining the positions for each attribute in any audio editing program such as Praat or
Audacity and manually entering the values into the “oto.ini.” This can be helpful when
creating an UTAUloid collaboratively, as tunings from different sources (such as multiple
collaborators working simultaneously on different computers) can be copied and pasted into
the same “oto.ini” file without issue or created jointly through online platforms like Google
Docs; it can also be useful if tuning a particular sample requires more detailed spectrogram
manipulation (e.g. window length) than the view within OpenUTAU provides.
With the audio samples and a finished “oto.ini” file in the same directory, the
UTAUloid is complete. Metadata about the UTAUloid (including the UTAUloid’s name,
voice donors, tuners, illustrators, contributors, and contact information) can be specified in
CJAL * RCLA Sleeper 72
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
text files called “readme.txt” and “character.txt,” which are read by UTAU software to
provide information in the application, and a picture can optionally be set for the voicebank
by including it as “image.png” in the same directory. The directory can then be compressed
(into a .zip file, for example) and shared; anyone with the OpenUTAU application (or other
UTAU software) installed can then easily load in the .zip file and use the enclosed
UTAUloid.
Of course, while massive collaboration is a key part of the general UTAUloid
experience, in many language revitalization contexts creators may want to restrict the usage
of an UTAUloid in accordance with community norms and wishes, and it is important to
note that the distribution of any UTAUloid is always completely at the discretion of its
creator. An UTAUloid could be kept for use only within a given community, rather than
distributed over the internet, or even not be shared at all – while UTAU allows for almost
limitless collaboration, there is no requirement or expectation that a voicebank be shared or
shared in any specific way.
Along with distribution, the acceptable usage of an UTAUloid can also be specified
with a license provided with the voicebank. General UTAU guidelines already prohibit the
use of any UTAUloid to advance racism or synthesize hate speech, among other
stipulations, but many voicebank licenses also specify restrictions prohibiting commercial
use, redistribution, or altering the tunings or voice samples, for example. In language
revitalization contexts, licenses are especially important in that they allow UTAUloids to
align with Indigenous practices of data sovereignty (Kukutai & Taylor, 2016). The open-
source nature of UTAU and its community of massive collaboration has important potential
for language revitalization, but as Brinklow et al. (2019) point out, “‘open source’ requires
a more nuanced application in the Indigenous context, especially at the interface between
‘tool’ and ‘data’” which UTAUloids inhabit (p. 405). Crafting an appropriate license that
grants access to the voicebank, the language data which it contains, and/or the music
created from it in accordance with a community’s customs and protocols is a vital part of
using UTAUloids in Indigenous language revitalization, and this process can in turn be a
fruitful part of wider community conversations around Indigenous data sovereignty and
licensing in general.
14
For this Cherokee UTAUloid, I included a brief, bespoke license (reproduced
below) based on the Cherokee concept of ᎦᏚᎩ (gadugi), an ethic of coming together in
cooperative community labor to work towards shared goals, represented by the image of
“building one fire.” As praxis, ᎦᏚᎩ is closely connected to language and cultural
preservation (Cushman, 2010), and so drawing on the Cherokee Nation’s practice of
making language resources – including in-person and online language classes, learning
materials, and digital archives of stories and songs (Cushman, 2013) – available to anyone
interested in learning Cherokee, including non-citizens, the license for this UTAUloid
stipulates that it can be used freely by anyone for the purposes of furthering the Cherokee
language:
This UTAU voicebank is distributed under a ᎦᏚᎩ (gadugi) license. ᎦᏚᎩ is the
Cherokee ethic of coming together to work towards a common goal.
In the spirit of ᎦᏚᎩ, this UTAUloid can be used freely by anyone working towards
our common goal of helping to revitalize the Cherokee language.
CJAL * RCLA Sleeper 73
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
Please do not alter, modify, or redistribute this voicebank elsewhere.
As with all UTAU voicebanks, it is prohibited to use ᎧᏃᎩᏍᏗ/Kanogisdi to
advance racism, sexism, or bigotry of any kind.
ᏩᏙ!
Results and Discussion
The result of the above method is a complete Cherokee UTAUloid, ᎧᏃᎩᏍᏗ
(kanogisdi, “singing”): an early release with much room for improvement, but which can
already be used to simply and easily create vocal melodies in the language. Once the
UTAUloid is loaded into OpenUTAU or other UTAU-compatible software, users place
musical pitches on the piano roll-style score, with vertical height indicating pitch and width
indicating length; clicking on a placed note allows users to specify the syllable to be sung,
based on the file labels and aliases used in the voicebank. Figure 10 shows a simple melody
composed and sung in Cherokee using ᎧᏃᎩᏍᏗ in OpenUTAU, which can be heard
alongside other samples at the project’s webpage:
15
Figure 10
Using the Cherokee UTAUloid
ᎧᏃᎩᏍᏗ
in OpenUTAU
CJAL * RCLA Sleeper 74
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
Initial goals for the ᎧᏃᎩᏍᏗ project include writing songs based on the lesson
dialogues from the Cherokee Nation self-paced online classes – both as a learning exercise
for my own language reclamation, and to help other learners through all the benefits music
can provide to language learning – as well as collaborations with other Cherokee artists to
eventually create a character model and key art for the UTAUloid.
As a tool for music creation, UTAUloids have several characteristics which make
them appealing to music producers. Many of these same features also help make them
uniquely useful as tools for language revitalization. The first of these features is that
UTAUloid is a very versatile format for music-making; melodies composed in the
framework are both portable and exportable in a variety of formats. The sung audio can be
exported into a .wav file, for instance, and the resulting “vocal track” can be added to any
digital audio workstation, including open-source systems such as Audacity or LMMS
(LMMS Developers, 2020), and combined with additional tracks to create a complete song.
UTAUloid scores can also be shared in their native UST format, which means that vocal
melodies composed or started by one user can be edited, tweaked, and/or finished by any
other user that they share the .ust file with. Finally, UTAUloid melodies can be exported
natively to MIDI, providing for easy importing into open-source graphical music editors
like MuseScore (MuseScore Contributors, 2023).
This (ex)portability makes UTAUloid particularly well-suited to collaborative work,
and that, combined with the culture of “massive collaboration” around UTAUloid, creates
unique potential in terms of Indigenous language revitalization. In the context of language
revitalization, massive collaboration means that speakers, learners, community members –
and depending on the context and license, potentially people outside of the community – of
all different language and musical skill levels can all participate collaboratively in the same
musical-linguistic project of creating and using UTAUloids. Speakers who may not
consider themselves ‘musical’ could contribute lyrics or voice samples; learners who may
not feel comfortable writing lyrical songs in the language could contribute melodies or take
on lyric-writing as part of the language learning process, perhaps in collaboration with
fluent speakers (see for example Ashton, 2020 on the Jersey Song Project); musicians could
contribute melodies or backing tracks without any specific knowledge of or interest in the
language; and other community members could contribute with any of the other elements
of UTAUloid's inherently “multidisciplinary creation” (Sousa, 2014), including art, writing,
and character and costume design. In the Cherokee case in particular, I see the concept of
massive collaboration as a reflection of ᎦᏚᎩ gadugi in action in a musicolinguistic
framework for language revitalization: coming together from different perspectives in the
shared work of language and community, engaging a variety of skillsets, motivations, and
lived experiences, and building “one fire from the flames and fuel of many fires”
(Cushman, 2010, p. 95).
The distributed model of creation associated with massive collaboration can also be
especially advantageous in languages with large, engaged digital diasporas, like Cherokee.
With around 10,000 speakers in the Cherokee Nation, but over 450,000 registered tribal
members spread out across the reservation, Oklahoma, the 23 recognized “At-Large”
communities throughout the United States, and beyond – many of whom are strongly
engaged with the Cherokee Nation and the language revival movement through the internet
and social media – this digital diaspora represents tremendous creative potential for projects
like UTAUloid. The massive collaboration inherent in making music with UTAUloid
CJAL * RCLA Sleeper 75
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
represents a particularly participatory and accessible way to engage the benefits of music
for language revival, as well as a unique medium to bring people together through
musicking, whether in physical community, or in the shared virtual space of an Indigenous
online creative community.
In addition to multidisciplinary collaboration, another of the primary reasons users
turn to UTAUloid in any context is also a key advantage for language revitalization: low-
resource music making. Provided someone has access to a given UTAUloid and a
computer, they can make vocal music in that language for free. Combining the vocal tracks
created by UTAUloid with free and open-source digital audio workstations (such as LMMS
or Audacity) and their integrated instrument synthesizers then allows users to create full,
commercial quality vocal songs without any monetary investment beyond a computer.
UTAUloid is also an example of truly low-resource music making in terms of physical
space; without the need for studio recording equipment or a low-noise environment, and
with the ability to compose start-to-finish with headphones, UTAUloid songs can be
composed in their entirety in any space with a computer – including quiet public spaces
such as libraries, and louder shared spaces like community centers.
Perhaps most importantly in terms of language revival, UTAUloid also represents
an excellent potential avenue to bolster youth involvement with language revitalization.
The UTAUloid community is overwhelmingly a community of youth; many popular
UTAUloids have been created by teenagers, as is the case for most of the minoritized
language UTAUloids mentioned above. And crucially, rather than introducing a new
concept to young people for the purposes of language revitalization, UTAUloid represents a
concept, platform, and community (through both UTAUloid and Vocaloid) that is already
incredibly popular with young people (Condry, 2011; Lam, 2016; Le, 2014; March, 2022;
Yin, 2018). Furthermore, the countries in which UTAUloid and Vocaloid are most popular –
Canada, the United States, Mexico, Japan, China, Taiwan, Malaysia, and Indonesia,
16
among others – are all countries in which colonialism and linguistic assimilationist policies
have led to widespread endangerment of, and subsequent revitalization efforts in support of,
Indigenous languages (Eberhard et al., 2022). As one example of this worldwide appeal,
Figure 11 shows a sold-out 2016 live concert by Hatsune Miku (projected via hologram and
accompanied by a live band) in San Francisco's 2,300-seat Warfield Theatre, packed
completely full of young people swinging glowsticks in time to the music – music
composed and created by their peers, fellow Vocaloid and UTAUloid fan-producers.
UTAUloid offers a way for young people to bring that energy to their own
revitalizing languages, and to bring those languages into new, exciting domains. This can in
turn be a powerful source of momentum in language revitalization; as the organizers of a
youth workshop designed to create new music in Pitjantjatjara observe: “if media content
created by young people is of the same high standard as other media with which they
engage, then their own language content will always be more popular – we have observed
this hands down” (Sometimes & Kelly, 2010, p. 88).
It is worth noting here that while Vocaloids and UTAUloids can be employed to
create any type of vocal music – and users have, for example, composed operas and
synthesized traditional folk songs with both – the electronic, computerized, and by design
not-fully “natural” nature of the language sounds they create means that UTAUloids may
not be culturally appropriate in all language or musical contexts, or compatible with
protocols around musicking and song creation within a given community.
CJAL * RCLA Sleeper 76
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
Figure 11
Hatsune Miku Concert in San Francisco, 2016 (photo by author)
Further, Indigenous language revitalization and music revitalization are interrelated (Brown
et al., 2017; Grant, 2014; Marett & Barwick, 2003), and in some cases where communities
are also working to protect or revitalize traditional musical practices, UTAUloids could
potentially be seen as orthogonal – or even harmful – to a community’s goals in cultural
reclamation.
In Cherokee culture, however, there is a longstanding custom of adapting new and
non-“traditional” musical genres into the Cherokee language, as part of a general cultural
practice of “[making] things Cherokee simply by doing them” (Snyder, 2016, p. 62). This
can be seen from the 19th century integration of hymn-singing, to more recent examples of
Cherokee-language reggae, heavy metal, pop, and hip-hop on the ᎠᏅᏛᏁᎵᏍᎩ
(anvdvnelisgi ‘performers’) compilation by Horton Records
17
(Eaton, 2022), young
campers at the Snowbird Cherokee Traditions Language Camp in North Carolina
performing a Cherokee translation of Lil NasX’s 2019 country trap hit Old Town Road
18
(Knoepp, 2019), immersion school students playing with language in a Cherokee version of
the Ghostbusters themesong (Snyder, 2016), and the experimental Cherokee-language
fusion of futuristic disco, electronica, and modern dance of Elisa Harkin’s ᎦᏬᏂᏍᎩ ᏦᎢ
(gawonisgi tsoi ‘Radio III’).
19
In the Cherokee context, then, the electronic sounds and
worlds afforded by UTAUloid represent an exciting new approach to explore, as well as a
continuation of this tradition of drawing new musical aesthetics and practices into the
Cherokee world.
CJAL * RCLA Sleeper 77
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
Finally, while UTAUloids in this context are primarily a vehicle for language
reclamation and revitalization, they also represent a unique point of fruitful collaboration
between applied linguistics and Indigenous language revitalization. Because of the vital
importance of language learning in Indigenous language revitalization, this collaboration is
often rightfully conceptualized in terms of language education (whether in classrooms or in
community) and second language acquisition. Here, applied linguistics can provide tools
for language learning, teacher training, and creative and efficient ways which people can
use to help learn, teach, and pass on ancestral and revitalizing languages, contextualized
within the ontologies, experience, and decolonization-focused social justice lens of
Indigenous language revitalization (e.g. Daniels & Sterzuk, 2022; McIvor, 2020).
What UTAUloids represent is an expansion of the applied linguistics knowledge
base which can be fruitfully brought to this collaboration – in the phonetic, phonotactic, and
practical recording skills required to create an UTAUloid. In this particular skillset, the
UTAUloid project echoes applied linguistic approaches which leverage phonetic methods
and computer technology for teaching L2 pronunciation (Chun, 1998; Chun & Jiang, 2022;
Levis, 2007). But here, rather than offering frameworks for language learning, applied
linguistics provides the background required to create a tool for musical language creation
(the UTAUloid), which can then be taken up by Indigenous language revitalization in many
different ways: as an individual project of language reclamation; as a community
collaboration, potentially engaging contributors with diverse skills across space and time in
one or many musical-linguistic projects; as an educational tool for songwriters or in
classroom or community language pedagogies; or simply as an instrument for people to
express themselves creatively in their ancestral language. The expansion of this relationship
between applied linguistics and Indigenous language revitalization into the realm of music
– with all of the distinctive benefits and affordances that that results in for learners and
speakers – also points to the potential of bringing musical disciplines working towards
language revitalization – like ethnomusicology (Ashton, 2022; Przybylski, 2018; Sparling
et al., 2022) – into the conversation as well.
Daniels and Sterzuk (2022) describe the relationship between applied linguistics and
Indigenous language revitalization in practical terms: “Applied linguistics exists to offer
real-life solutions to language problems. We see threats to Indigenous languages as the most
urgent language problem of our time and therefore understand that applied linguistics is
called upon to offer solutions” (p. 15). Because of that urgency, and especially because of
the demanding and potentially overwhelming nature of language revitalization work
(Walsh, 2018), the intersection of applied linguistics and Indigenous language revitalization
should ideally offer a variety of solutions that are innovative, exciting, meaningful for
participants independent of larger-scale outcomes or metrics like speaker numbers, and –
crucially – joyful; all of which, I believe, describes UTAUloid.
Conclusion
This study has aimed to introduce and explain the creation of a Cherokee
UTAUloid, as an example of one way in which applied linguistics can contribute directly,
using specialist skills, to the musical side of Indigenous language revitalization. Through
their focus on “massive collaboration” – allowing speakers, learners, musicians, and others
to work together on the same project – low-resource music production (in terms of both
cost and space), and youth involvement (through existing popularity in global youth
CJAL * RCLA Sleeper 78
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
culture), UTAUloids are uniquely well-situated to serve a variety of roles in language
revitalization. And along with all the potential benefits from music made with UTAUloids
– such as attracting learners and engaging speakers, creating opportunities for speech
community, and supporting language learning – even the act of creating an UTAUloid itself
allows speakers and learners who may not consider themselves “musical” to contribute to
language revitalization through music, to engage in meaningful language reclamation
resulting in tangible tools for others to take up, and to explore a unique and fruitful
confluence of applied linguistics and Indigenous language revitalization.
Acknowledgements
My heartfelt thanks to Matt Gordon, Eric Campbell, Tim Cooley, Marianne Mithun,
Heather Sparling, Kit Ashton, and all of the participants at the A’ Chànain Cheòlmhor:
Language Revitalization through Music workshop for their inspiring conversations and
helpful feedback on earlier versions of this work. I’m also deeply grateful to Ed Fields for
sharing his knowledge of the Cherokee language with me and countless others — ᏩᏙ! —
and to the UTAUloid community for sharing their virtual voices and creative energy with
the world.
Correspondence should be addressed to: Morgan Sleeper
Email: msleeper@macalester.edu
Notes:
1
The symbol <v> in Cherokee romanization represents a nasalized schwa /ə/
2
Samples of various Vocaloids can be heard at <https://soundcloud.com/cryptonfuturemeida>
3
Examples of recent Hatsune Miku songs can be found at <https://soundcloud.com/hatsunemikuofficial> (for
audio) and <https://www.youtube.com/@HatsuneMiku> (for videos)
4
Formally, the software framework itself is known as 'UTAU', and the derived synthesizers are known as
'UTAUloids', but 'UTAUloid' is used here as a convenient term for the combined concept, on analogy with
‘Vocaloid’.
5
Available at <http://utau2008.xrea.jp/>
6
Available at <http://utau-synth.com/>
7
Available at <http://www.openutau.com>
8
One important exception to this is that UTAU usage guidelines forbid the creation of a voicebank from a
speaker without their express consent. This includes (specifically) deceased persons, which means that
archival recordings of speakers who have passed away should not serve as voicebank sources.
9
From oto (音), meaning 'sound' in Japanese
10
There are also standalone programs designed specifically for UTAU tuning; one popular option is vLabeler
(available at <https://vlabeler.com>), which will also create and update the ‘oto.ini’ file automatically.
11
Also referred to as ‘fixed’ in other UTAU software
12
Also referred to as ‘blank’
13
The length of the consonant region can optionally be manipulated in OpenUTAU for stylistic effect on a
per-note basis with the ‘VEL’ (velocity) parameter.
14
As one example, the Kaitiakitanga License developed by Māori media organization Te Hiku Media (2023)
is based on the Māori concept of kaitiakitanga (loosely, ‘guardianship’) rather than ownership of data, and
was created to protect against the digital colonization of Māori data and knowledge.
15
<https://github.com/morgansleeper/Kanogisdi>
16
These countries have all hosted the Hastune Miku Expo Vocaloid concert series – a concert tour presented
by Crypton Future Media featuring their Vocaloids singing as holographs, accompanied by live band
17
<https://hortonrecords.bandcamp.com/album/anvdvnelisgi-performers>
CJAL * RCLA Sleeper 79
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
18
<https://soundcloud.com/tsalagiseli/old-town-road-cherokee-version>
19
<https://elisaharkins.bandcamp.com/album/radio-iii>
CJAL * RCLA Sleeper 80
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
References
Abrate, J. H. (1983). Pedagogical applications of the French popular song in the foreign
language classroom. The Modern Language Journal, 67(1), 8–12.
https://doi.org/10.1111/j.1540-4781.1983.tb02492.x
Antoine, J. (2015). The role of traditional songs in the maintenance and preservation of
Lakota language and culture. In N. Ostler, & B. W. Lintinger (Eds.), Proceedings of
foundation for endangered languages (Vol. 19, pp. 17–23).
Archer, J. (2002, October 4). Poi-E. New Zealand Folk Song. https://folksong.org.nz/poi_e/
Ashton, K. (2020). The Jersey song project. In J. Olko, & J. Sallabank (Eds.), Revitalizing
endangered languages: A practical guide (pp. 289–290). Cambridge University
Press.
Ashton, K. (2022). Êcliaithe Man Tchoeu [Light Up My Heart]: Applied ethnomusicology
and the revitalisation of the endangered language of Jèrriais [PhD Thesis].
Goldsmiths, University of London, England, U.K.
Audacity Team. (2022). Audacity®: Free Audio Editor and Recorder (3.2.5) [Computer
software]. https://www.audacityteam.org/
Barrett, R. (2016). Mayan language revitalization, hip hop, and ethnic identity in
Guatemala. Language & Communication, 47, 144–153.
https://doi.org/10.1016/j.langcom.2015.08.005
BBC News. (2018, October 20). Gwenno “sparks record numbers” in Cornish exams. BBC
News. https://www.bbc.com/news/uk-england-cornwall-45917661
Berez-Kroeker, A. L., Gabber, S., & Slayton, A. (2023). Recent Advances in Technologies
for Resource Creation and Mobilization in Language Documentation. Annual
Review of Linguistics, 9,195–214. https://doi.org/10.1146/annurev-linguistics-
031220-120504
Boersma, P., & Weenink, D. (2023). Praat: Doing phonetics by computer (6.3.09)
[Computer software]. http://www.praat.org/
Bracknell, C. (2020). Rebuilding as research: Noongar song, language and ways of
knowing. Journal of Australian Studies, 44(2), 210–223.
https://doi.org/10.1080/14443058.2020.1746380
Bracknell, C., Bracknell, K., Fenty Studham, S., & Fereday, L. (2021). Supporting the
performance of Noongar language in Hecate. Theatre, Dance and Performance
Training, 12(3), 377–395. https://doi.org/10.1080/19443927.2021.1943506
Brinklow, N. T., Littell, P., Lothian, D., Pine, A., & Souter, H. (2019). Indigenous language
technologies & language reclamation in Canada. In European Language Resources
Association (Ed.), Language technology for all (LT4All): Enabling language
diversity and multilingualism worldwide (pp. 402–406). UNESCO Publishing.
Brown, R., Manmurulu, D., Manmurulu, J., O’Keeffe, I., & Singer, R. (2017). Maintaining
song traditions and languages together at Warruwi (western Arnhem Land). In J. W.
Wafer, & M. Turpin (Eds.), Recirculating songs: Revitalising the singing practices
of indigenous Australia. Hunter Press.
Caldwell, E. (2023, March 27). Can a newly installed cellphone tower help preserve a
language? NPR. https://www.npr.org/2023/03/27/1166180027/can-a-newly-
installed-cell-phone-tower-help-preserve-a-language
Cherokee Nation. (2022). Cherokee national youth choir. Cherokee Nation Education
Services. https://www.cherokee.org/all-services/education-services/youth-
CJAL * RCLA Sleeper 81
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
leadership/cherokee-national-youth-choir/
Cherokee Phoenix Staff. (2022, August 9). Cherokee Nation launches innovative speaker
services program. Cherokee Phoenix
https://www.cherokeephoenix.org/services/cherokee-nation-launches-innovative-
speaker-services-program/article_dcd76e94-17e7-11ed-9509-ffd3a6635251.html
Chun, D. M. (1998). Signal analysis software for teaching discourse intonation. Language
Learning & Technology, 2(1), 74–93. http://dx.doi.org/10125/25033
Chun, D. M., & Jiang, Y. (2022). Using technology to explore L2 pronunciation. In J. M.
Levis, T. M. Derwing, & S. Sonsaat-Hegelheimer (Eds.), Second language
pronunciation: Bridging the gap between research and teaching (pp. 129–150).
John Wiley & Sons.
Clusternote. (2014). Vocal synthesizer piano roll—Sakura Sakura.jpg [Digital image].
https://commons.wikimedia.org/wiki/File:Vocal_synthesizer_piano_roll_-
Sakura_Sakura.jpg
Condry, I. (2011, July 11). Miku: Japan’s virtual idol and media platform. MIT Center for
Civic Media. https://civic.mit.edu/index.html%3Fp=1749.html
Cotter, C. (2001). Continuity and vitality: Expanding domains through Irish-language radio.
In L. Hinton, & K. Hale (Eds.), The green book of language revitalization in
practice (pp. 301–316). Brill.
Cru, J. (2018). Micro-level language planning and YouTube comments: Destigmatising
indigenous languages through rap music. Current Issues in Language Planning,
19(4), 434–452. https://doi.org/10.1080/14664208.2018.1468960
Crypton Future Media. (n.d.). Who is Hatsune Miku? Retrieved March 31, 2023, from
https://ec.crypton.co.jp/pages/prod/virtualsinger/cv01_us
Cushman, E. (2010). Gadugi: Where the fire burns (still). In S. Kahn, & J. Lee (Eds.),
Activism and rhetoric: Theories and contexts for political engagement (pp. 56–61).
Routledge.
Cushman, E. (2012). The Cherokee syllabary: Writing the people’s perseverance.
University of Oklahoma Press.
Cushman, E. (2013). Wampum, Sequoyan, and story: Decolonizing the digital archive.
College English, 76(2), 115–135. http://www.jstor.org/stable/24238145
Daniels, B., & Sterzuk, A. (2022). Indigenous language revitalization and applied
linguistics: Conceptualizing an ethical space of engagement between academic
fields. Canadian Journal of Applied Linguistics, 25(1), 1–18.
https://doi.org/10.37213/cjal.2022.31841
Danos, D., & Turin, M. (2021). Living language, resurgent radio: A survey of Indigenous
language broadcasting initiatives. Living Language, 15, 75-152
http://hdl.handle.net/10125/24971
Davis, G. M. (2017). Songs in the young learner classroom: A critical review of evidence.
ELT Journal, 71(4), 445-455. https://doi.org/10.1093/elt/ccw097
Dołowy-Rybińska, N. (2020). Fest-noz and the revitalisation of the Breton language. In J.
Olko, & J. Sallabank (Eds.), Revitalizing endangered languages: A practical guide
(pp. 287–288). Cambridge University Press.
Eaton, K. (2022, April 19). Cherokee artists are preserving their language through music.
Good Good Good. https://www.goodgoodgood.co/articles/cherokee-language-
music-album
Eberhard, D. M., Simons, G. F., & Fennig, C. D. (Eds.). (2022). Ethnologue: Languages of
CJAL * RCLA Sleeper 82
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
the world (25th ed.). SIL International. http://www.ethnologue.com
Engh, D. (2013). Why use music in English language learning? A survey of the literature.
English Language Teaching, 6(2), 113-127. https://doi.org/10.5539/elt.v6n2p113
Golla, V. (2010). North America. In C. Moseley (Ed.), Encyclopedia of the world’s
endangered languages (pp. 1–96). Routledge.
Good, A. J., Russo, F. A., & Sullivan, J. (2015). The efficacy of singing in foreign-language
learning. Psychology of Music, 43(5), 627–640.
Goodyear, S. (Director). (2019, May 1). Meet the N.S. teenager who sang Blackbird by The
Beatles entirely in Mi’kmaw. CBC Radio. https://www.cbc.ca/radio/asithappens/as-
it-happens-wednesday-edition-1.5118294/meet-the-n-s-teenager-who-sang-
blackbird-by-the-beatles-entirely-in-mi-kmaw-1.5118296
Grant, C. (2014). Music endangerment: How language maintenance can help. Oxford
University Press.
Hinton, L. (2001). Language Revitalization: An Overview. In L. Hinton, & K. Hale (Eds.),
The green book of language revitalization in practice (pp. 3–18). Academy Press.
Jadii. (2009). Sachi Eika UTAUloid. http://terraloid.tumblr.com/sachi_eika
Johnson, H. (2012). “The group from the west”: Song, endangered language and sonic
activism on Guernsey. Journal of Marine and Island Cultures, 1(2), 99–112.
https://doi.org/10.1016/j.imic.2012.11.006
Jolly, Y. S. (1975). The use of songs in teaching foreign languages. The Modern Language
Journal, 59(1/2), 11–14. https://doi.org/10.2307/325440
Kenmochi, H. (2010). VOCALOID and Hatsune Miku phenomenon in Japan. Proceedings
of InterSing 2010: First Interdisciplinary Workshop on Singing Voice, 1–4.
https://www.isca-archive.org/intersinging_2010/kenmochi10_intersinging.pdf
Kenmochi, H., & Ohshita, H. (2007). VOCALOID - Commercial singing synthesizer based
on sample concatenation. Proceedings of Interspeech 2007, 4009–4010.
https://www.isca-archive.org/interspeech_2007/kenmochi07_interspeech.pdf
Knoepp, L. (2019, September 24). Heading down the
‘
old town road
’
to teach the Cherokee
language. Blue Ridge Public Radio. https://www.bpr.org/news/2019-09-24/heading-
down-the-old-town-road-to-teach-the-cherokee-language
Kukutai, T., & Taylor, J. (2016). Data sovereignty for Indigenous peoples: Current practice
and future needs. In T. Kukutai, & J. Taylor (Eds.), Indigenous data sovereignty:
Toward an agenda (pp. 1–22). Australian National University Press.
Lam, K. Y. (2016). The Hatsune Miku phenomenon: More than a virtual J-pop diva. The
Journal of Popular Culture, 49(5), 1107–1124. https://doi.org/10.1111/jpcu.12455
Le, L. K. (2014). Examining the rise of Hatsune Miku: The first international virtual idol.
The UCI Undergraduate Research Journal, 1-12.
Leonard, W. Y. (2012). Reframing language reclamation programmes for everybody’s
empowerment. Gender and Language, 6(2), 339–367.
https://doi.org/10.1558/genl.v6i2.339
Levis, J. (2007). Computer technology in teaching and researching pronunciation. Annual
Review of Applied Linguistics, 27, 184–202.
https://doi.org/10.1017/S0267190508070098
Llewellyn, M. (2000). Popular music in the Welsh language and the affirmation of youth
identities. Popular Music, 19(3), 319–339.
https://doi.org/10.1017/S0261143000000192
CJAL * RCLA Sleeper 83
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
LMMS Developers. (2020). LMMS (1.2.2) [Computer software]. https://lmms.io/
Lucas, O. R. (2021). Kaitiakitanga, Whai Wāhi and alien weaponry: Indigenous
frameworks for understanding language, identity and international success in the
case of a Māori metal band. Popular Music, 40(2), 263–280.
https://doi.org/10.1017/S0261143021000131
MacKinnon, K. (2005). Cornish/Kernewek. In D. O’Néill (Ed.), Rebuilding the Celtic
languages: Reversing language shift in the Celtic countries (pp. 214–274). Y Lolfa.
March, L. (2022). “Wrap you up in my blue hair”: Vocaloid, Hyperpop, and Identity in
“Ashnikko Feat. Hatsune Miku – Daisy 2.0.” Television & New Media, 24(8), 894-
910. https://doi.org/10.1177/15274764221093599
Marett, A., & Barwick, L. (2003). Endangered songs and endangered languages. In J.
Blythe, & R. M. Brown (Eds.), Maintaining the links: Language identity and the
land (pp. 144–151). Foundation for Endangered Languages.
McCarty, T. L., Nicholas, S. E., Chew, K. A. B., Diaz, N. G., Leonard, W. Y., & White, L.
(2018). Hear our languages, hear our voices: Storywork as theory and praxis in
Indigenous-language reclamation. Daedalus, 147(2), 160–172.
https://doi.org/10.1162/DAED_a_00499
McIvor, O. (2020). Indigenous language revitalization and applied linguistics: Parallel
histories, shared futures? Annual Review of Applied Linguistics, 40, 78–96.
https://doi.org/10.1017/S0267190520000094
Mithun, M. (1999). The languages of native North America. Cambridge University Press.
Montgomery-Anderson, B. (2015). Cherokee reference grammar. University of Oklahoma
Press.
Moseley, C. (Ed.). (2010). Atlas of the world
’
s languages in danger. UNESCO Publishing.
MuseScore Contributors. (2023). MuseScore (4.0.2) [Computer software].
https://musescore.org/
Nummelin, G. (2020). One song, many voices: Revitalising Ainu through music. In J. Olko
& J. Sallabank (Eds.), Revitalizing endangered languages: A practical guide (pp.
291–293). Cambridge University Press.
Peter, L., Hirata-Edds, T., Feeling, D., Kirk, W., Mackey, R. “Wahde,” & Duncan, P. T.
(2017). The Cherokee Nation immersion school as a translanguaging space. Journal
of American Indian Education, 56(1), 5-31.
https://doi.org/10.5749/jamerindieduc.56.1.0005
Przybylski, L. (2018). Bilingual hip hop from community to classroom and back: A study in
decolonial applied ethnomusicology. Ethnomusicology, 62(3), 375–402.
https://doi.org/10.5406/ethnomusicology.62.3.0375
Przybylski, L. (2021). Indigenizing the mainstream: Music festivals and Indigenous popular
music. IASPM Journal, 11(2), 5–21. https://doi.org/10.5429/2079-
3871(2021)v11i2.2en
Pulte, W., & Feeling, D. (1975). Outline of Cherokee grammar. In W. Pulte (Ed.),
Cherokee-English dictionary (pp. 235–355). Cherokee Nation of Oklahoma.
Salazar, J., Belmar, G., Scanlon, C., Troiani, G., & Campbell, E. W. (2021). Bridging
diaspora: Technology in the service of the revitalization of Sàꞌán Sàvǐ ñà Yukúnanǐ
(Mixtec). Endangered Languages and Diaspora - XXV Annual Conference
Proceedings, 176–185.
https://www.linguistics.ucsb.edu/sites/default/files/sitefiles/people/campbell/Salazar
CJAL * RCLA Sleeper 84
Canadian Journal of Applied Linguistics: Special Issue, 27, 2 (2024): 52-84
etal-bridgingdiaspora-2021_11_04.pdf
Schmidt, J. (2003). German rap music in the classroom. Die Unterrichtspraxis/Teaching
German, 36(1), 1-14. https://doi.org/10.2307/3531679
Sheehan, M. (2016). Mana Wahine: Māori women in music. Te Kaharoa, 9(1), 76-90.
https://doi.org/10.24135/tekaharoa.v9i1.12
Snyder, S. L. (2016). Poetics, performance, and translation in Eastern Cherokee language
revitalization [PhD Thesis]. Columbia University, New York, New York, U.S.
Sometimes, B., & Kelly, A. (2010). Ngapartji Ngapartji: Indigenous language in the arts. In
J. Hobson, K. Lowe, S. Poetsch, & M. Walsh (Eds.), Re-awakening languages:
Theory and practice in the revitalisation of Australia’s Indigenous languages (pp.
85–89). Sydney University Press.
Sousa, A. M. D. de. (2014). A colaboraão massiva de Hatsune Miku: Software vocaloid
como catalisador de criaões colectivas, grassroots e multidisciplinares na
subcultura otaku [The massive collaboration of Hatsune Miku: Vocaloid software as
a catalyst for collective, grassroots and multidisciplinary creation in otaku
subculture]. Revista Croma, Estudos Artísticos, 2(3), 121–137.
http://hdl.handle.net/10451/12237
Sparling, H., MacIntyre, P., & Baker, S. (2022). Motivating traditional musicians to learn a
heritage language in Gaelic Nova Scotia. Ethnomusicology, 66(1), 157–181.
https://doi.org/10.5406/21567417.66.1.09
Te Hiku Media. (2023). Kaitiakitanga License.
https://github.com/TeHikuMedia/Kaitiakitanga-License/blob/tumu/LICENSE.md
Tegge, F. (2018). Pop songs in the classroom: Time-filler or teaching tool? ELT Journal,
72(3), 274–284. https://doi.org/10.1093/elt/ccx071
Tuttle, S. G. (2012). Language and music in the songs of Minto, Alaska. Language
Documentation and Description, 10, 82–112. https://doi.org/10.25894/ldd191
Tuttle, S. G., & Lundström, H. (2015). Taking charge: Learner agency in the transmission
of song and speech traditions. In N. Ostler, & B. W. Lintinger (Eds.), Proceedings of
Foundation for Endangered Languages (Vol. 19, pp. 38–44).
Uchihara, H. (2016). Tone and accent in Oklahoma Cherokee (1st ed.). Oxford University
Press.
UTAU Wikia contributors. (n.d.). Sachi Eika. UTAU Wikia. Retrieved March 31, 2023,
from http://utau.wikia.com/wiki/Sachi_Eika
Vallejo, J. M. (2019). Revitalising language through music: A case study of music and
culturally grounded pedagogy in two Kanien’ke:ha (Mohawk) language immersion
programmes. Ethnomusicology Forum, 28(1), 89–117.
https://doi.org/10.1080/17411912.2019.1641124
Walsh, M. (2018). “Language is like food...” Links between language revitalization and
health and well-being. In L. Hinton, L. M. Huss, & G. Roche, The Routledge
handbook of language revitalization (pp. 5–12). Routledge New York.
https://doi.org/10.4324/9781315561271
Yin, Y. (2018). Vocaloid in China: Cosmopolitan music, cultural expression, and multilayer
identity. Global Media and China, 3(1), 51–66.
https://doi.org/10.1177/2059436418778600
ᎠᎾᏗᏍᎪᎢ/Anadisgoi. (2023, February 21). Cherokee Nation reaches 450,000 Cherokee
citizens. ᎠᎾᏗᏍᎪᎢ/Anadisgoi. https://anadisgoi.com/index.php/government-
stories/cherokee-nation-reaches-450-000-cherokee-citizens
