Content uploaded by Daniel W. Hieber
Author content
All content in this area was uploaded by Daniel W. Hieber on Nov 28, 2019
Content may be subject to copyright.
Vol. 9 (2015), pp. 344–350
http://nflrc.hawaii.edu/ldc
http://hdl.handle.net/10125/24669
Mari C. Jones. 2014. Endangered languages and new technologies. Cam-
bridge: Cambridge University Press. 228 pp. ISBN: 978-1107049598. Price
$99. Also available as e-Book ($79).
Reviewed by Daniel W. Hieber,University of California, Santa Barbara
The perceived relationship between technology and minority languages has changed dras-
tically over the years. Linguists previously viewed technologies like television and radio
as major drivers of language shift—veritable ‘cultural nerve gas’ (Krauss 1992:6). Today,
by contrast, the role of technology is viewed primarily as a positive, enabling one. The
existence of the present volume demonstrates just how far those views have shifted. The
fundamental premise of this book is that, “The ever-increasing availability of new tech-
nologies, from visual to aural archiving to digitization of textual resources and electronic
mapping, have the potential to revolutionize the documentation, analysis and revitalization
of endangered languages for the linguist and indigenous community alike” (xiii). Of course,
the editor and the collective authors do not view technology as a cure-all for language shift,
and are appropriately cautionary in their suggestions for the application of technology. But
the central message of this book is a hopeful one: the suite of ever-cheaper and ever-higher-
quality tools available today can, with appropriate sensitivity to the cultural contexts in
which they are applied and to their congruence with community goals and resources, be a
powerful tool in the reemergence and flourishing of minority languages.
This hopeful message shines through most prominently in Nicholas Ostler’s excellent
introduction, ‘Endangered languages in the New Multilingual Order per genus et differen-
tiam.’ Ostler argues that the forces which previously gave mega-languages—and especially
English—their status are becoming less relevant: “Since linguistic dominance of this kind
is always based on past social dominance (military, economic, cultural, or religious), and
the social factors that favoured English-speakers over others are losing force, there is scope
for change in the multilingual order of the world” (1). In support of this view, Ostler notes
that the regions of the world which have shown the most rapid economic advancement this
century are precisely the non-English speaking ones, and that even English’s predominance
on the internet is waning. Moreover, as automated translation becomes more robust, choice
of language will become more a matter of preference than necessity. Ostler sees the poten-
tial for what he calls the New Multilingual Order, a world where, “English will carry on as
a useful lingua franca, a support mechanism, but one that will be increasingly unnecessary.
However, the direction of flow […] is increasingly towards a world where choice of lan-
guage will express its inherited position and felt loyalties within the human race, even to
quite small groups. In this world of aspiration, all will speak as they like, and yet the world
will understand them” (12–13).
This idea of technology as the great leveler is one of several themes that recur throughout
the book. In Tjeerd de Graaf, Cor van der Meer, and Lysbeth Jongbloed-Faber’s chapter,
“The use of new technologies in the preservation of an endangered language: The case of
Frisian,” they report that the cheap cost of distance-learning technologies has given Frisian
language learners access to qualified language instructors, whereas prior to the prevalence of
Licensed under Creative Commons
Attribution-NonCommercial 4.0 International
E-ISSN 1934-5275
Review of Endangered language and new technologies 345
these technologies, hiring a language instructor would have been financially impractical due
to the small number of pupils able to attend class in any given locality. Distance-learning
technologies have enabled Frisian language instructors to take advantage of economies of
scale, making paid language instruction economically viable (143).
In general, tools developed for well-resourced majority languages are often utilizable
by minority language communities, who then do not need to undertake the fixed costs of
development. The more prevalent the use of the tool among majority languages, the greater
the infrastructure for it becomes, the lower its costs, and the easier it is for minority lan-
guages to utilize. Matt Coler and Petr Homola’s chapter, “Rule-based machine translation
for Aymara,” is an excellent illustration of this process at work. As they note, even though
the methods of machine translation were originally developed for well-studied languages,
“there was no intrinsic reason why this should remain their focus in the long term” (10). The
internet provides a repository of linguistic data on a rapidly growing number of languages,
and all of this publicly-facing data is open to analysis. As Ostler notes in his introduction,
“The net effect will be that the smaller 99 percent of the world’s languages […] will have
a corresponding opportunity to become accessible; the ‘long tail’ need no longer be disre-
garded” (2–3).
The adoption of existing technologies by minority languages is also an example of a
broader pattern that emerges repeatedly throughout the book: the repurposing of a technol-
ogy for uses other than its original intended ones. Anthony Scott Warren & Geraint Jennings
report exactly this process in their chapter, “Allant contre vent et mathée’: Jèrrais in the
twenty-first century.” They describe how Les Pages Jèrrais, a website originally launched
with the sole intention of making Jèrrais language material publicly available, came to func-
tion as a linguistic corpus for tracking usage, lexical variation, and neologisms, and also as
a primitive spell-checker (136). Another kind of seldom-discussed repurposing that many
linguists encounter is the repurposing of legacy archival materials for modern pedagogi-
cal materials or linguistic analysis. Jeffrey E. Davis makes this point especially salient
in his chapter, “American Indian Sign Language: Documentary linguistic methodologies
and technologies,” where legacy video recordings constitute a significant component of his
work to document and create pedagogical materials for American Indian Sign Language
(173). Once these recordings were digitized, it was possible to conduct extensive lexical
comparisons between legacy and modern sources. Thus, even legacy material in obsolete
formats might still be repurposed for ends none of their creators might have imagined; one
never knows what uses the data or tools you create might have. This fact, of course, should
encourage caution in planning for the longevity of documentary materials, and being extra
attentive to details of access rights and permissions.
Other notable examples of repurposing in this volume include: the reuse of a keyboard
layout created by native Me'phaa writers to create keyboards for different operating systems
(since the layout itself is a technological solution to a particular problem, and thus a type
of technology) (50); the use of 6,000 tweets by fifty Frisian-speaking adults to create the
beginnings of a Frisian spell-checker (149); and a set of Frisian-language DVDs which
originally had limited distribution, but made their way onto YouTube, and are now used as
pedagogical materials by teachers (143).
Other themes thread their way through the book as well, which is impressive given
that the chapters themselves vary widely in their content and focus, and indeed in their
very conception of ‘technology.’ With chapters on keyboards, corpus creation, machine-
translation, data longevity and archiving, and many other topics, a potential reader could be
forgiven for assuming that the volume lacks cohesion. But the strong contribution this book
Language Documentation & Conservation Vol. 9, 2015
Review of Endangered language and new technologies 346
makes to language documentation and revitalization is, in my opinion, not in the specifics
of the individual projects outlined in the chapters themselves. That is not to say that the
projects and ideas outlined in each contribution are not extremely valuable in their own right;
it simply acknowledges the fact that not every documentary linguist is in a situation where
the types of projects and technologies outlined in these chapters apply. Rather, what makes
this book worth reading regardless of the relevance of the particular projects to the reader’s
own is the wealth of information and advice on design principles, project planning, practical
and useful goal-setting, best practices in community collaboration, and accommodating
community needs and cultural preferences. In the remainder of this review, I take up a few
of the commonalities exhibited across different chapters, and what they have to teach us
about the use of technology in language documentation and revitalization. Some of these
tendencies I see as positive models to emulate, while others I offer as constructive criticism,
but all of them provide valuable lessons for any project where technology plays an integral
role.
The first theme is the authors’ very conception of technology itself. The term ‘tech-
nology’ is actually quite vague, and could refer to everything from hardware to search al-
gorithms to data formats or user interfaces, or many, many other things. Despite this, the
contributors focused centrally on technology as realized through tools, that is, the particu-
lar programs, applications, or websites that assist users in performing certain tasks or op-
erations. This is a decidedly end-user perspective on technology, which asks, ‘What does
technology allow me to do?’ and it makes sense that linguists and community members,
who are usually not themselves technologists, would be most concerned with this perspec-
tive. Technicalities and details of implementation are generally glossed over. In Aiméé
Lahaussois’ chapter “The Kiranti comparable corpus: A prototype corpus for the compar-
ison of Kiranti languages and mythology,” for example, the author explicitly states that,
“what is advocated here is not a particular software configuration but, rather, a concept, the
technical implementation of which could be realized in a number of different ways” (17–18).
This is exemplary of the approach taken by the majority of the authors in this book.
While a purely conceptual understanding a given technological tool is useful, it is at the
same time a bit unfortunate because linguists do ultimately have to confront the nitty-gritty
details of implementation as well. To give just one example, few of the authors discuss user
interfaces in any detail. However, a well-designed user interface can sometimes make all the
difference between adoption of a tool or apathy towards it by researchers and community
members. Hugh Paterson III makes this point extremely well in his chapter, “Keyboard
layouts: Lessons from Me'phaa and Sochiapam Chinantec designs,” and it is worth quoting
at length:
When language documenters and linguists build digital solutions such as key-
board layouts, they need to bear in mind that these products may have lasting
effects on communities. As service providers, they have ethical and profes-
sional obligations to seek out not only solutions but great solutions. […] When
linguistic and technical expertise is offered to communities of endangered lan-
guage speakers and writers, we need to not only design solutions, we need also
to offer well-designed solutions. Just because something is usable and useful
does not mean it is desirable. When a speech community does not want to use a
given input method (keyboard layout), the response should not be: ‘Well, they
simply don’t want it enough.’ Keyboard layouts are not just products, they are
experiences (54).
Language Documentation & Conservation Vol. 9, 2015
Review of Endangered language and new technologies 347
Paterson goes on to exemplify this attention to usability with a helpful overview of design
challenges in creating keyboards for minority languages.
Lahaussois’ chapter on the Kiranti corpus, mentioned above, is also a good example of
how writers can engage with and share details of implementation in a way that is useful to
other researchers/revitalizers without being overly meticulous. It includes screenshots and
descriptions of different interfaces in their database management tool, such as concordanc-
ing and side-by-side views of parallel/comparable texts. For an audience primarily focused
on what tools allow them to do, this exposition of the tool is extremely useful, because
it allows potential adopters to see precisely what they can do if they were to apply that
technology to their own projects.
In general, however, the authors are not especially concerned to relate the implemen-
tational details behind their projects. Even Lahaussois’ discussion of the Kiranti corpus,
for example, does not tell us anything about what the programming world calls ‘the stack,’
or the suite of programming languages, operating systems, browsers, servers, etc. that are
necessary to build and maintain a given product. Describing a software stack need not be
tedious. Even a cursory overview can be extremely valuable for other project teams who
might want to emulate a particular technological framework. Was the tool created with
nothing more than JavaScript, a modern web browser, some coffee and a weekend, or did it
require extensive server-side scripting and expertise in security and authentication? Does it
use a data format on the backend that is already compatible with one’s own data, or would
using this tool require extensive manipulation and reformatting of that data? Knowledge of
these facts can help project teams make more informed decisions about what is practical for
them to achieve given the resources and expertise available to them, and helps them locate
individuals with the proper skill sets for their project. Lahaussois for instance notes that for
their project, “The goal of setting up a distributed network of databases accessible via the
Internet proved too ambitious” (45). What we do not know is what tools they considered
and what made those tools untenable. Similarly, Paterson notes, “The literature offers rela-
tively little in terms of guiding principles for designers of keyboard layouts. The absence is
not completely unexpected since human-computer interaction such as keyboarding is often
treated and discussed as a sub-discipline of computer science or psychology […] rather than
of linguistics” (52). Thus the documentary and revitalization community as a whole would
do well not to shy away from the technological details of implementation for the tools they
use and create.
Moving away from constructive criticisms, one laudable theme that emerges from this
volume is a focus on interoperability, i.e., the ability for tools to work with other tools,
or work across different languages. An excellent example of interoperability at work is
Sjief Barbiers’ chapter “European Dialect Syntax: Towards an infrastructure for documen-
tation and research of endangered dialects.” This is the only chapter to discuss not just a
project infrastructure, but also a broader collaborative research infrastructure, “where lin-
guists can store and access the relevant data and where they can cooperate in the description
and analysis of these data” (35). Barbiers goes on to provide some helpful implementational
details illustrating a good principle for collaborative research teams to follow: decentraliza-
tion of data management accompanied by centralization of search capabilities, allowing
different research groups to operate independently of each other, but share their data: “the
databases and tools included in such an infrastructure should not be stored on one central
server. Rather, they should constitute a distributed network of databases, searchable using
a common search engine (preferably via the Internet) and analysable with using [sic] a car-
tographic tool in order to visualize the geographical distribution of one or more syntactic
Language Documentation & Conservation Vol. 9, 2015
Review of Endangered language and new technologies 348
properties. The advantage of such a decentralized infrastructure is that every research group
involved is able to maintain and update their own database independently” (43). The result
is that each researcher’s data is interoperable and transferable with that of others.
Interoperability is also one of the eight fundamental questions for endangered language
technology projects discussed in Russell Hugo’s chapter, “Endangered languages, technol-
ogy and learning: Immediate applications and long-term considerations,” a well-considered
chapter with some excellent advice for planning the technological components of any lan-
guage revitalization project. Hugo states that a primary consideration should be the avoid-
ance of “content lock,” defined as follows: “if a solution is designed and content is integrated—
as much content and organization as possible should be able to be extracted and easily mi-
grated to a future platform” (102). Interoperability therefore includes not just considerations
of transfers between different tools, but different times as well.
The focus on interoperability does unfortunately also suffer somewhat from a lack of
attention to implementational details. For example, while many authors advocate storing
and transferring data in XML, this alone is not enough for interoperability, precisely be-
cause XML schemas can be implemented in many ways. The transcription tool ELAN
(Max Planck Institute for Psycholinguistics; cf. Brugman & Russell 2004) formats its data
in XML in a drastically different way than does the glossing and lexicography tool FLEx
(SIL 2015). Moreover, XML, far from being a “future standard format” (102) as it was orig-
inally hailed, is being largely abandoned in favor of the far simpler format known as JSON,
which is used to store and exchange data in most web-based applications today. JSON is
perfectly designed for exchanging highly structured data like linguistic texts, something
XML was not primarily designed to do. This means that the majority of linguistic tools,
including those presented in this book, are not easily interoperable with most modern web
technologies, and this is unfortunate because the future of technology most definitely lies
with the web.
But Hugo here has excellent advice as well: “Complicated software development is
arguably a less than ideal use of resources. Rather than seeking to ‘reinvent the wheel’ for
each endangered language, it may be worth looking around to see whether applications that
have already been created can also aid the documentation, development and distribution of
learning materials for endangered language efforts” (110). Utilizing pre-existing tools is
an excellent way to ensure interoperability between one’s own project and others using that
tool, and also helps encourage the expansion of that tool’s infrastructure by putting it to new
uses.
One of the project outputs mentioned in some fashion or another by every contributor
is searchability. Digital technologies provide a variety of new means of searching one’s
data at different levels. Some of the searchable features demonstrated in the present vol-
ume include language-internal variation, geographic variation, syntactic constituency, or
even basic searches on glosses, words, or part-of-speech tags. Searchability also enables
the application of big data techniques to smaller corpora. Searchability is ultimately what
distinguishes data from archival materials. This point is made especially clear in Bernard
Bel and Médéric Gasquet-Cyrus’s chapter, “Digital curation and event-driven methods at
the service of endangered languages.” They caution against outputs that amount to “little
more than a huge and widely disseminated showcase, which is hardly useful for revitaliza-
tion” (114). Mere digitization is not enough. The authors sagely point out that we don’t
want to wind up with the digital version of shoeboxes of fileslips sitting in our closets: “if
the old tapes are merely replaced by digital recordings stored on personal computers or
unconnected websites, has there been any real change?” (87).
Language Documentation & Conservation Vol. 9, 2015
Review of Endangered language and new technologies 349
Searchability comes with some attendant difficulties as well, namely, questions of how
one searches for material across different languages, or across inconsistent glossing conven-
tions. The impulse of the contributors and many other linguists is towards standardization.
Barbiers, for example, decrying the fact that “different research groups/language areas tend
to use distinct [part of speech] tags, which makes it impossible to search multiple databases
using a single set of tags,” argues that, “A common, standardized and well-defined tag set is
therefore essential” (44). Dorothee Beermann’s chapter “Data management and analysis for
endangered languages,” however, offers another approach. She discusses a software tool
called TypeCraft, used for annotation of textual data. This includes morphological glossing
of a number of different languages, each with their own traditions of grammatical termi-
nology and analysis. How does one reconcile these varied analyses in a way that makes
the corpus consistently searchable? TypeCraft solves this problem by dividing data into
two classes: common data, which are standardized across corpora (such as ISO codes or
glosses from the GOLD ontology), and individual data, which comprise user-defined cate-
gories. This seems a better solution than insisting on complete standardization of glosses,
an approach advocated by other authors in the volume. The crosslinguistic application of
grammatical categories is fundamentally an issue of linguistics and not data structures, and
sits at the heart of many a debate in typology and linguistic theory. Data structures should
not impose theoretical constraints on the data, and forcing standardization of glossing con-
ventions does just that.
The story is different for metadata, however. As Barbiers also notes, “It is important that
every database be enriched with standardized metadata so that the database can be selected
on the basis of its properties. These metadata can include, for example, information on
the language area and the dialects, dates of the recordings and profiles of consultants” (43–
44). Since metadata are not dependent on the linguist’s analysis or linguistic theory in the
same way that morphological glosses are, and since metadata have the primary function of
locating and identifying items in a collection, standardization of metadata should be strongly
encouraged.
Another theme that features prominently in this volume is open access of both data
and tools, while remaining sensitive to issues of access and permission. In fact, a central
point in Beermann’s chapter on data management is that recent technologies in some ways
make managing access easier than ever before, given the ease of setting up user groups
and profiles and restricting access to users in permitted groups (81). And Cecilia Odé, in
her chapter on ‘Language documentation and description from the native speaker’s point of
view: The case of Tundra Yukaghir,’ shows just how beneficial such open-access tools can
be. She relates the launch of a free access, interactive e-learning module about language
shift focusing primarily on Tundra Yukaghir of Siberia and Mpur of West Papua, and how
its broad availability has made it a valuable tool for raising awareness about language shift.
Moreover, speakers of other minority languages easily related the film to their own language
context and recognized the situation of the Tundra Yukaghir as analogous to their own,
fostering fruitful discussions regarding language vitality, documentation, and revitalization
for their own communities.
The final and perhaps most important thread that runs throughout this book is the way
in which technology both furthers and in some cases makes possible increased community-
academic collaboration and community involvement. Paterson summarizes this nicely:
“The global levelling of information access through the Internet also enables speakers of
endangered languages and academics to engage more fully with each other—rather than, as
before, operating in different social circles. Roles such as ‘linguist’, ‘language documenter’
Language Documentation & Conservation Vol. 9, 2015
Review of Endangered language and new technologies 350
or ‘endangered language speaker’, which might previously have been mutually exclusive,
can therefore now be fulfilled by ‘academics’ and ‘native speakers’ alike” (50). And collab-
orative spaces are key to productive language learning, as noted by Hugo when discussing
what makes sound pedagogy for language revitalization. He explains that, “Technology
may provide additional learning time via online courses, spaces to collaborate and commu-
nicate at a distance” (98).
Taken together, the themes in this volume lend credence to Ostler’s positive outlook
for minority languages. The authors exemplify the way in which new technologies and the
tools that stem from them can be, when appropriately leveraged, the great leveler, putting
minority languages on equal footing with more dominant ones. They also show how mi-
nority language communities can co-opt tools first created for larger, more well-resourced
languages, thus reducing the cost of adoption, and fostering innovative ways of leveraging
preexisting tools and data for new purposes. They show how the utility of such tools is
greatly enhanced by following principles of open access and interoperability, and how this
in turn fosters a greater degree of collaboration and crossover between academics and com-
munity members. And in purely practical terms, every one of the authors demonstrates how
much there is to be gained from enabling even simple searching across digitized data and
metadata. Add to this the range of helpful advice for project planning, technology design,
and strategies for maximizing adoption and use of revitalization tools sprinkled through-
out the book, and the result is a volume that members of any language documentation or
revitalization project would do well to read.
References
Brugman, Hennie & Albert Russell. 2004. Annotating multimedia/multi-modal resources
with ELAN. In Maria Teresa Lino, Maria Francisca Xavier, Fátima Ferreira, Rute Costa
& Raquel Silva (eds.), Proceedings of LREC 2004, 4th International Conference on Lan-
guage Resources & Evaluation, 2065–2068. Paris: European Language Resources Asso-
ciation. http://www.lrec-conf.org/proceedings/lrec2004/.
Krauss, Michael. 1992. The world’s languages in crisis. Language 68(1). 4–10.
Max Planck Institute for Psycholinguistics. ELAN. Nijmegan, The Netherlands: The Lan-
guage Archive. http://tla.mpi.nl/tools/tla-tools/elan/.
SIL. 2015. Fieldworks Language Explorer (FLEx). http://fieldworks.sil.org/.
Daniel W. Hieber
dhieber@umail.ucsb.edu
Language Documentation & Conservation Vol. 9, 2015