Volume 5 Issue 1 (2007)
DOI:10.1349/PS1.1537-0852.A.314
Note: Linguistic Discovery uses Unicode characters
to represent phonetic symbols. Please see Optimizing Display
for requirements to accurately reproduce this page.
Ngbugu digital wordlist: A test case for best practices in archiving
and presenting language
documentation[1]
Gary F. Simons
SIL International
Kenneth S. Olson
SIL International and University of North Dakota
Paul S. Frank
SIL International
Language documentation faces challenges of data preservation and
accessibility. Data can be lost due to physical deterioration (e.g. field notes
or tape recordings) or outdated format (e.g. Microsoft Word 3.0). Archived data
is typically difficult to access, and it is sometimes found that the archived
information is inadequate for research purposes. Increased interest in language
documentation has coincided with advancements in digital technologies, offering
hope for meeting these challenges. This paper discusses the archiving of a
204-item wordlist of Ngbugu, an Ubangian language spoken in Central African
Republic, employing best practice recommendations. Our solution includes: TIFF
digital imaging of the original handwritten transcription, WAV digital recording
of the wordlist, descriptive markup encoding of the wordlist in XML employing
Unicode transcription, viewing and playback via an XSLT style sheet that renders
the information in HTML, publishing metadata for resource discovery with the
Open Language Archives Community (OLAC), and depositing the original materials
and digital representations in an institutional archive committed to long-term
preservation and access.
1.
Introduction
Significant strides have been made recently in the
documentation of the world’s languages, but along with this have arisen
certain challenges. Much of this documentation remains unpublished and is
therefore inaccessible to others. Some of that documentation, e.g. audio tape
recordings, will eventually be lost due to physical deterioration. Materials can
be easily converted into digital form to increase accessibility, but as Bird and
Simons (2003) observe, unless steps are taken to ensure its longevity,
“much digital language documentation and description becomes inaccessible
within a decade of its creation” (p. 557).
These challenges point to the need for principled approaches to making
language documentation long lasting, accessible, and re-usable. In responding to
this need, linguists have begun to concern themselves with the development of
best practices for digital language documentation. The Open Language Archives
Community (OLAC) [http://www.language-archives.org]
exists for this purpose, and the concern is also a driving force behind the
E-MELD project (Electronic Metastructure for Endangered Language Data) [http://www.emeld.org].
This paper discusses the digital archiving of a 204-item wordlist in
Ngbugu, an Ubangian language spoken in Central African Republic by approximately
95,000 people (ISO 639–3 code: [lnl], Gordon 2005, [http://www.ethnologue.com/show_language.asp?code=lnl]).
The process employed is a test case for various best practice recommendations
concerning the archival documentation of language resources, especially those of
Bird and Simons (2003), Plichta and Kornbluh (2002), and MATRIX (2001). Many of
these best practice recommendations have been elaborated at the E-MELD School of
Best Practices in Digital Language Documentation [http://www.emeld.org/school]. The paper
builds on the process laid out in Frank and Simons (2003). We report the results
of our project to prepare these materials for long-term archiving (the
“archival form”) and to provide present-day access of the
information via the Internet (the “presentation form”) (Simons
2006).
It may be helpful to begin by showing the final results. Examples (1)
and (2) provide links to the presentation form and the archival form,
respectively, of the Ngbugu digital wordlist.
(1) Presentation form (Olson and Mbomate 2007)
Ngbugu
wordlist as a web page, with integrated metadata, images, and
recordings
(2) Archival form (Olson 2006)
OLAC
metadata record (3 kilobyte XML file)
Wordlist
with time-aligned transcriptions (48 kilobyte XML file)
Digitized original audio recording (82 megabyte WAV file)
Digital images of original wordlist forms (2 TIFF files of 8 megabytes
each)
Definitions of XML (Extensible Markup Language), WAV, TIFF
(Tagged Image File Format), and other pertinent terms employed here are
available at [http://emeld.org/school/glossary.html].
First, click on the link in (1) to see the form that has been developed
for publishing the Ngbugu wordlist as an interactive web page using
today’s presentation technologies. On that page, click on the links in
The Wordlist section to see the resource description (i.e. metadata) and
the images of the original transcriptions. Then click on a loud speaker icon to
hear the pronunciation of the word as recorded. The icon is linked to a WAV
file; thus your web browser will attempt to play it with the sound program that
is set up as the default WAV player on your computer.
After exploring all the features of the presentation form, click on the
two links in (2) to see the XML form of the metadata and the XML form of the
wordlist from which the presentation form was generated. The other two elements
in the archival form (the complete recording and the images of the field
transcriptions) are too large to make available via the web medium at this time.
The complete set of results is available on a CD-ROM from the SIL Language and
Culture Archives, 7500 W. Camp Wisdom Rd., Dallas, TX 75236–5699, USA,
archive_dallas@sil.org.
We give an overview of our approach in section 2. In section 3, we
enumerate the best practice recommendations we followed. Section 4 presents an
overview of the process we followed to convert the original materials into
archival and presentation forms. Finally, in section 5, we offer concluding
remarks.
2. Solution
The original Ngbugu wordlist materials collected in Central
African Republic included two items: a two-page wordlist form and a 16-minute
recording on audio cassette. The wordlist form presented the standardized
wordlist of 204 items from Moñino (1988). For each item the form provided
a prompt in French and a space for the transcription of the elicited form. In
this case, the form was filled in with handwritten Ngbugu orthographic
transcriptions by Jacques Vermond Mbomate, a Ngbugu speaker literate both in
Ngbugu and French. Some items included a suggested alternative pronunciation in
parentheses, or an indication of uncertainty concerning the data. The second
author verified the list in consultation with Mr. Mbomate, and together they
produced a transcription employing the International Phonetic Alphabet (IPA
1999).
The second author then created a revised list in Microsoft Word for
Windows 2000 that included the French prompt, the orthographic rendering, and
the IPA rendering. The accompanying audio cassette contained a recording of Mr.
Mbomate repeating the revised list. He produced the French prompt first, then
the Ngbugu equivalent. The recording was made with a Marantz PMD 420 monaural
cassette recorder and an Audio-Technica ATM 33a microphone. The recording
session took place on March 6, 2004, at the ACATBA center (l’Association
Centrafricaine pour la Traduction de la Bible et d’Alphabétisation)
in Bangui, Central African Republic.
As Simons (2006) observes, a linguist must do two things to ensure that
language documentation will persist far into the future. First, in order to
ensure that the materials will still be readable from the software point of
view, they must be put into a file format that software of the future will still
be able to interpret. Second, in order to ensure that the materials will still
be readable from the hardware point of view, they must be deposited with an
institutional archive that will ensure that they are migrated as needed to fresh
media lest they perish on media that become obsolete (e.g. 5 ¼" diskettes)
or unreadable (e.g. the limited shelf-life of CD-Rs, cf. Byers 2003).
In order to meet the first objective of putting the material into a
format appropriate for long-term archiving, we set out to do the
following:
• Make archive-quality digital images of the original handwritten
wordlist forms,
• Make an archive-quality digital version of the audio
recording,
• Create an archive-quality digital encoding of the transcribed
wordlist as an XML file that encodes French and English glosses of each word,
the transcription in Ngbugu orthography and IPA (using Unicode for the
encoding), any additional notes, and the start and stop times in the digital
audio files for each Ngbugu utterance, and
• Create a metadata description of this set of archival
materials.
In order to meet the second objective of ensuring the
on-going availability of the materials far in the future beyond the life of the
current hardware and media, our plan was to:
• Place all of these materials in the SIL Language and Culture
Archives for long-term preservation.
But long-term access is not our only objective. We also want
the materials to be available today in an easy-to-access form on the Internet.
In order to meet this objective, we also set out to:
• Generate versions of the digital audio and digital images that
are suitable for browsing on the web,
• Develop an XSLT (Extensible Stylesheet Language) script that
generates an HTML (HyperText Markup Language) presentation form for viewing the
digitally encoded wordlist that provides links to display the digital images of
the original transcriptions and to play back the recording of each
utterance,
• Publish the presentation form of these materials on the Internet
to enable linguists to inspect them, and
• Make these materials known through OLAC using their metadata
standards.
3. Best practice
guidelines
In this project, we were guided by best practice
recommendations for digitizing the audio recordings (Plichta and Kornbluh 2002),
for digitizing the images of the transcription (MATRIX 2001), and for creating
digital language documentation and description in general (Bird and Simons
2003). The following tables summarize relevant aspects of these recommendations
and indicate the degree to which this project was able to adhere to these
recommendations.
Table 1 addresses guidelines for digitizing audio recordings. Plichta
and Kornbluh (2002) recommend a sample rate of 96,000 Hz and a bit depth of 24
bits as a standard for archive-quality digital audio, but also note that 44,100
Hz, 16-bit is adequate for technical purposes. Nearly all acoustic information
pertinent to language is below 11,000 Hz, and the upper limit of hearing for
most people is 22,000 Hz (Ladefoged 2003:18, 26). Since the highest frequency
that can be reconstructed from a digital recording is half the sampling rate
(aka “the Nyquist frequency”, cf. Nyquist 1928, Shannon 1949), the
44,100 Hz rate is sufficient for speech and hearing. Since our digitizing
equipment was not adequate at the time for the higher sampling rate and bit
depth, we opted for the lower resolution. (See the EMELD glossary for an
explanation of the terms “sample rate” and “bit depth”
[http://emeld.org/school/glossary.html].)[2]
We avoided the use of minidisc recorders and MP3 files, because these
involve compression techniques that result in loss of sound quality. This is of
particular concern for those wishing to perform acoustic analysis of the
recordings, who would want the data to be as unprocessed as possible.
Table 1: Recommendations for digital audio (Plichta and
Kornbluh 2002)
Recommended Best Practice
|
Ngbugu Wordlist Project
|
Recommended for archival purposes
-sample rate: 96,000 Hz
-bit depth: 24-bit
|
Lack of appropriate hardware prevented us from following these
recommendations.
|
Sufficient for technical purposes
-sample rate: 44,100 Hz
- bit depth: 16-bit
|
This is the standard that we followed.
|
Oversampling delta-sigma A/D converter with dither added prior to
sampling
|
Lack of appropriate hardware prevented us from following this
recommendation.
|
WAV file format
|
This is the format that we used.
|
Table 2 addresses guidelines for digitizing images of
textual materials. MATRIX (2001) proposes separate recommendations for master
images and for access images. We followed the former recommendations for
generating the archival form and the latter for the presentation form.
Table 2: Recommandations for digital imaging (MATRIX
2001)
|
Recommended Best Practice
|
Ngbugu Wordlist Project
|
Master images
|
|
|
Bit depth
|
8-bit grayscale or 24-bit color
|
8-bit grayscale
|
Scanning resolution
|
300 dpi for original documents if smaller than 11" × 17", 200 dpi
if larger than 11" × 17"
|
300 dpi
|
Image size
|
Size of original document at scan resolution
|
Original image size of 8.5" × 11" is preserved.
|
Format
|
Uncompressed TIFF
|
Uncompressed TIFF
|
Access images
|
|
|
Bit-depth
|
8-bit grayscale or 24-bit color
|
8-bit grayscale
|
Scanning resolution
|
72–90 dpi depending on character height
|
72 dpi
|
Image size
|
Original size, at 72–90 dpi
|
Original image size of 8.5" × 11" is preserved.
|
Format
|
For documents smaller than 8.5" × 14": 4-bit interlaced GIF for
8-bit grayscale images or 8-bit interlaced GIF for 24-bit color images
For documents larger than 8.5" × 14": 8-bit greyscale JPEG for
grayscale images or 24-bit color JPEG, RGB mode for color images.
|
8-bit interlaced GIF
|
Table 3 addresses the problem of the portability of digital
language resources. This includes the synchronic portability of resources across
a multiplicity of present-day computing platforms as well as the diachronic
portability of today’s resources to the computing platforms of the future.
Bird and Simons (2003) identify seven dimensions of portability and propose best
practice guidelines designed to maximize the ability of digital language
resources to move across different computing platforms and to remain usable far
into the future. Two of the seven dimensions—citation and
preservation—are mostly relevant to the institutions that publish
and archive resources. The other five, however—content,
format, discovery, access, and rights—are
particularly relevant to the linguists who create language resources. The
following table repeats some of Bird and Simons’ key recommendations in
these five areas and describes how the current project has responded.
Table 3: Recommendations regarding the portability of
language documentation and description (Bird and Simons
2003)
Recommended Best Practice
|
Ngbugu Wordlist Project
|
Content
|
|
When texts are transcribed, provide the primary recording (without
segmenting it into clips)
|
The full elicitation is provided in a 16-minute WAV file.
|
Transcriptions should be time-aligned to the underlying recording in
order to facilitate verification.
|
Each response is time-aligned.
|
Format
|
|
Use open formats supported by multiple software vendors.
|
The formats used for archival forms are open: XML (for transcription),
WAV (for audio), TIFF (for images).
|
Use Unicode for character encoding.
|
IPA transcriptions are encoded in Unicode.
|
Use a descriptive markup system (preferably XML) for textual
information.
|
The whole wordlist (including glosses, transcriptions, and time
alignments) is represented in an XML file with descriptive markup
tags.
|
Provide a human-readable version of the same information in a suitable
presentation format.
|
The XML wordlist is transformed to HTML for presentation in a web
browser.
|
Discovery
|
|
Describe the resource using the metadata standard of OLAC.
|
An OLAC-conformant resource description is supplied.
|
Make the resource known to the world at large by publishing the
metadata description with an OLAC data provider.
|
The metadata are published via the OLAC data provider for SIL’s
Language and Culture Archive.
|
Access
|
|
Make resources accessible to all interested users.
|
Presentation form is published on the web.
|
Publish in such a way that users can access the original materials to
manipulate them in novel ways.
|
Archival form of full resources is available by ordering a
CD-ROM.
|
Rights
|
|
Make a clear statement of terms of use so that users know what they may
do with the material.
|
The resource description states the materials are copyrighted and
available to all under standard terms of Fair Use.
|
Identify and protect any sensitivities inherent in the
material.
|
There are no known sensitivities and this is stated in the resource
description.
|
4. Results and
process
Following Simons (2006), we distinguish three forms of data
in this project:
• Working form: the form in which information is stored as it is
created and edited,
• Archival form: the form in which information is stored for
access long into the future, and
• Presentation form: the form in which information is presented to
the public.
Part of the design of this project is to distinguish
archival and presentation formats for the information. In this section we
describe these two forms of the data in detail. In addition, we describe certain
aspects of the processes we used to create them.
4.1 Archival
form
Following the various best practice recommendations, the
archival form for the digital files are in open formats—XML for the
textual data (using the UTF-8 encoding for the Unicode character set), WAV for
the audio data, and uncompressed TIFF for the graphic images of the original
written documents. Open formats have greater longevity and can be transformed
into presentation formats that are more “reader friendly.”
The XML files are derived from the XML output of the TableTrans software
(Bird, et al. 2002) that was used to organize the French and English glosses,
Ngbugu orthographic and IPA transcriptions, and WAV file time-alignment data.
Figure 1 shows entries 14–16 in the XML file, including the item number,
French and English glosses, start and stop times for the Ngbugu utterance in the
master sound file, and the two transcriptions of the utterance. Where there are
two alternate Ngbugu words or pronunciations for a given prompt, these are
recorded in separate “response” blocks of XML data within a single
XML <item> element, as seen in item 15.
Figure 1: Entries 14–16 in the XML archive file for the
Ngbugu data
|
<item n=" 14">
|
|
|
|
|
<gloss xml:lang="fr"> bouche</gloss>
|
|
|
|
<gloss xml:lang="en"> mouth</gloss>
|
|
|
|
<response>
|
|
|
|
|
<audio start="51.775000" end="52.325000"/>
|
|
|
|
<orth> ma</orth>
|
|
|
|
<form> mà</form>
|
|
|
</response>
|
|
|
</item>
|
|
|
|
<item n=" 15">
|
|
|
|
|
<gloss xml:lang="fr"> bras/main</gloss>
|
|
|
|
<gloss xml:lang="en"> arm/hand</gloss>
|
|
|
|
<response>
|
|
|
|
|
<audio start="54.900000" end="55.575000"/>
|
|
|
|
<orth> könô</orth>
|
|
|
|
<form> könó </form>
|
|
|
|
<note>arm</note>
|
|
|
</response>
|
|
|
|
<response>
|
|
|
|
|
<audio start="57.075000" end="57.675000"/>
|
|
|
|
<orth> tchâneû</orth>
|
|
|
|
<form>
tʃánə́</form>
|
|
|
|
<note>hand</note>
|
|
|
</response>
|
|
|
</item>
|
|
|
|
<item n=" 16">
|
|
|
|
|
<gloss xml:lang="fr"> brouillard</gloss>
|
|
|
|
<gloss xml:lang="en"> mist; fog</gloss>
|
|
|
|
<response>
|
|
|
|
|
<audio start="59.575000" end="60.225000"/>
|
|
|
|
<orth> ndrö</orth>
|
|
|
|
<form>
ndʁō</form>
|
|
|
</response>
|
|
|
</item>
|
|
|
The XML file contains the information from the wordlist as a
structured plain text file. It uses the UTF-8 encoding of the Unicode character
set. Software capable of displaying data in UTF-8 should be able to render the
phonetic data faithfully, given a Unicode font that includes the IPA block of
characters. We used the Doulos SIL font in this project, as it conforms to our
specifications [http://scripts.sil.org/DoulosSILfont].
One problem that arises given the present state of computing concerns
the proper rendering of diacritics. In order for diacritics to be properly
placed when using Unicode characters, both the font and the software need to
have “smart font” capabilities, e.g. using OpenType or Graphite
technology [http://scripts.sil.org/RenderingGraphite].
In our data, this concerns the placement of acute, macron, and grave diacritics
(used to mark tone in the IPA) above a character with a dot (such as
“i”), and the stacking of a tone mark above a tilde (used to
indicate nasalization). The diacritics may not always be displayed properly; for
example, they may be superimposed rather than stacked. This is a problem that
should go away as future versions of software will likely incorporate these
capabilities. Despite this short-term problem, we still consider it best to opt
for Unicode encoding of the IPA characters rather than using a custom font. This
is because the underlying data is preserved for long-term storage following a
standard that is true to the original transcription, even if the rendering on
some systems may be less than ideal at present.
4.2 Presentation
form
Clearly, a presentation form of this data set is also
needed, and we have chosen to prepare a form of these data for web presentation.
For this purpose, we need three things: an XSLT style sheet to render the XML
file in HTML, individual WAV files for each of the utterances that are linked to
the data for playback, and GIF versions of the scanned images of the original
transcription. The rationale for creating individual WAV files and employing GIF
files for the presentation form is that they are much smaller in size than the
archival WAV and TIFF files. This is necessitated by present-day limitations of
storage for computers and especially bandwidth for Internet access. In the
future, these limitations will likely cease to exist, in which case the archival
standards could be employed for the presentation form. A sample of the HTML
presentation form of the first ten words in the wordlist is given in Figure
2.
Figure 2: Presentation form of the first ten words in the
Ngbugu wordlist
1.
|
abeille
|
bee
|
wräto
|
[wʁātò]
|
|
2
|
acide (vb)
|
tart
|
kpï,
kï
|
[kpī]
|
|
3.
|
aile
|
wing
|
mbrö
|
[mbʁō]
|
(adjective)
|
4.
|
aller
|
go
|
e
|
[ʔè]
|
|
5.
|
amer (vb)
|
be bitter
|
chü
|
[ʃū]
|
(adjective)
|
6.
|
animal
|
animal
|
gia
|
[già]
|
|
7.
|
année
|
year
|
ngû
|
[ngú]
|
|
8.
|
appeler
|
call
|
e
tchô
|
[ʔè
tʃó]
|
|
9.
|
arbre
|
tree
|
yö
|
[jō]
|
|
10.
|
attacher; lier
|
attach
|
i
(reu)
|
[ʔì]
|
|
4.3 The process
A major aspect of the Ngbugu project was linking the
transcription to the audio material. In the past, this would have been a
cumbersome process requiring finding each utterance and noting its start and
stop times in the master audio file using conventional audio software. However,
several programs are now available that make this process easier. These include
Transcriber [http://trans.sourceforge.net] and ELAN
[http://www.mpi.nl/tools/elan.html],
among others. We chose to use the TableTrans program (Bird, et al. 2002) from
the Linguistic Data Consortium [http://www.ldc.upenn.edu]. It allows the
linguist to follow a “play cursor” through a graphic representation
of the waveform as the recording is played back. The linguist may stop the
playback at any point, use the mouse to select a region of the wave, and then
enter annotations about that particular bit of the recording into a table with
user-defined fields.
In our application, we defined fields for the following six annotations:
the item number, the French prompt, the English translation, the orthographic
transcription of the Ngbugu utterance, the IPA transcription of the Ngbugu
utterance, and any additional notes that were entered on the original wordlist
forms. TableTrans is able to import the annotation data from a standard
comma-delimited file. We were able to exploit this feature since the annotations
were already available to us in a Microsoft Excel for Windows 2000 spreadsheet;
we saved that in comma-delimited format and then used TableTrans to add the time
alignment to the original recording. The fully annotated and time-aligned data
set can then be exported from the program in either comma-delimited (CSV) or XML
format. We then used an XSLT script to transform the XML “annotation
graph” output of TableTrans into the descriptive wordlist format shown in
Figure 1. Additionally, TableTrans automatically created the individual sound
files that correspond to each of the segments that have been identified in the
transcription process. (This functionality was ultimately the reason we chose to
use TableTrans, as it was not available in the other programs.) These are used
in the web-based presentation form of the data so that the playback of a single
utterance involves only the download of a small WAV file.
In this project, the sample rate of the recordings in both the archival
form and the presentation form was the same. However, it is likely that
researchers archiving data at a higher sampling rate (e.g. 96,000 Hz) will want
to convert the recordings to a lower sample rate (e.g. 44,100 Hz) for the
presentation form. When doing so, it is important to apply a low-pass pre-filter
to the data to remove all frequencies above the Nyquist frequency in order to
prevent aliasing (Ladefoged 1996:139–140). Some audio processing programs
(e.g. Cool Edit 2000) contain filters to do this.
4.4 Archive depositing and
resource discovery
Included with the archival form was a metadata record of
that resource. We followed OLAC’s standard for the format of this metadata
[http://www.language-archives.org/OLAC/metadata.html],
simply using a standard text editor to create the record. Alternatively, the
metadata record could be created using the OLAC Repository Editor [http://www.emeld.org/tools/ore.cfm].
The metadata was then verified using OLAC’s metadata validation service
[http://www.language-archives.org/tools/metadata/freestanding.html].
We deposited a CD-ROM of the archival form of the Ngbugu digital
wordlist (including the metadata record) with the SIL Language and Culture
Archives, located in Dallas, Texas. This institutional archive was then
responsible for entering the resource into its own metadata catalog and
publishing a record of the resource on the Internet [http://www.ethnologue.com/show_work.asp?id=47050].
Since the archive is a data provider to OLAC, the resource should appear on
pertinent searches done with OLAC search engines: [http://linguistlist.org/olac/] and [http://www.ldc.upenn.edu/olac/search.php]
5. Discussion
One success of this project was in preparing the same data
in both archival and presentation formats. The presentation format of the
textual data can be automatically generated from the archival form, thus
avoiding the need to maintain two distinct data sets. The archival data is
primary: without care in the preparation of the archival form of the data there
is a high likelihood that the information would be unusable within just a few
years because of changes in technology. With the current approach, it is
possible to have multiple scripts for generating multiple presentation formats.
While the archival format stays constant over time, future generations can
generate new presentation formats to take advantage of advances in presentation
technology.
Transcriptions and recordings exist for many other languages. The
TableTrans files prepared in this project could serve as a template for
preparing similar electronic data sets for these other languages. The new
transcriptions can be entered for a given language, the appropriate sound file
associated with it, and the time-alignment done. Other tools developed for this
current project could be easily adapted to facilitate the preparation of
presentation forms of wordlists for these other languages.
In addition, the general principles underlying the development of the
electronic version of the Ngbugu wordlist could be applied to other types of
language documentation, especially the distinction between archival and
presentation formats, the use of XML and Unicode for the textual data, and the
time-alignment of the audio information and textual information.
References
Bird, Steven, and Gary F. Simons. 2003. Seven dimensions of
portability for language documentation and description. Language
79/3.557–582. Preprint available at [http://www.ldc.upenn.edu/sb/home/papers/0204020/0204020-revised.pdf].
Bird, Steven; Kazuaki Maeda; Xiaoyi Ma; Haejoong Lee; Beth Randall;
and Salim Zayat. 2002. TableTrans, MultiTrans, InterTrans and TreeTrans: Diverse
tools built on the Annotation Graph Toolkit. Proceedings of the Third
International Conference on Language Resources and Evaluation. Paris: European
Language Resources Association. Online. URL: [http://arxiv.org/abs/cs/0204006].
TableTrans and the other related tools can be downloaded by selecting
“AGTK Windows” from [http://agtk.sourceforge.net/].
Byers, Fred R. 2003. Care and handling of CDs and DVDs: A guide for
librarians and archivists. (National Institute of Standards and Technology
Special Publication No. 500–252). Gaithersburg, MD: National Institute of
Standards and Technology/Washington, D.C.: Council on Library and Information
Resources. Online. URL: [http://www.itl.nist.gov/iad/894.05/docs/CDandDVDCareandHandlingGuide.pdf].
Casey, Mike & Bruce Gordon. 2007. Sound directions: Best
practices for audio preservation. Bloomington, IN & Cambridge, MA: Indiana
University & Harvard University. Online. URL: [http://www.dlib.indiana.edu/projects/sounddirections/papersPresent/sd_bp_07.pdf].
(Accessed 17 December 2007.)
Frank, Paul S., and Gary F. Simons. 2003. Sáliba wordlists:
A test case for best practices in archival documentation of an endangered
language. Paper presented at the Society for the Study of the Indigenous
Languages of the Americas annual meeting, Atlanta, Georgia, January 2–5,
2003.
Gordon, Raymond G. (ed.) 2005. Ethnologue: Languages of the world.
15th edition. Dallas: SIL. [http://www.ethnologue.com].
IASA-TC03. 2005. The safeguarding of the audio heritage: Ethics,
principles and preservation strategy, version 3. Online. URL: [http://www.iasa-web.org/IASA_TC03/TC03_English.pdf].
(Accessed 19 November 2007.)
IASA-TC04. 2004. Guidelines on the production and preservation of
digital audio objects. Aarhus, Denmark: International Association of Sound and
Audiovisual Archives.
International Phonetic Association. 1999. Handbook of the
International Phonetic Association. Cambridge: Cambridge University
Press.
Ladefoged, Peter. 1996. Elements of acoustic phonetics. Second
edition. Chicago: University of Chicago Press.
Ladefoged, Peter. 2003. Phonetic data analysis: An introduction to
fieldwork and instrumental techniques. Oxford: Blackwell.
MATRIX: The Center for Humane Arts, Letters and Social Sciences
Online at Michigan State University. 2001. Digital imaging for archival
preservation and online presentation: Best practices. ms. Online. URL: [http://www.historicalvoices.org/papers/image_digitization2.pdf].
Moñino, Yves (ed.) 1988. Lexique comparatif des langues
oubanguienne. Paris: Geuthner.
Nyquist, Harry. 1928. Certain topics in telegraph transmission
theory. Transactions of the American Institute of Electrical Engineers
47.617–644.
Olson, Kenneth S. 2006. Ngbugu digital wordlist: Archival form.
SIL-LCA-47050. SIL Language and Culture Archives, Dallas, Texas.
Olson, Kenneth S., and Jacques Vermond Mbomate. 2007. Ngbugu
digital wordlist: Presentation form. Linguistic Discovery 5/1:40-47 Online. URL:
[http://journals.dartmouth.edu/cgi-bin/WebObjects/Journals.woa/2/xmlpage/1/article/317].
Plichta, Bartek,
and Mark Kornbluh. 2002. Digitizing speech recordings for archival purposes. ms.
Online. URL: [http://www.historicalvoices.org/papers/audio_digitization.pdf].
Pohlmann, Ken C. 2006. Measurement and evaluation of
analog-to-digital converters used in the long term preservation of audio
recordings. Paper presentat at the "Issues in Digital Audio Preservation
Planning and Management" roundtable discussion, Washington, DC, March 10-11,
2006. Online. URL: [http://www.clir.org/activities/details/AD-Converters-Pohlmann.pdf].
(Accessed 17 December 2007.)
Shannon, Claude E. 1949. Communication in the presence of noise.
Proceedings of the Institute of Radio Engineers 37/1.10–21.
Simons, Gary F. 2006. Ensuring that digital data last: The priority
of archival form over working form and presentation form. SIL Electronic Working
Papers 2006–003. Dallas, TX: SIL International. Online. URL: [http://www.sil.org/silewp/2006/003/SILEWP2006-003.htm].
5 May 2006.
Authors’ contact information:
Gary F. Simons
SIL
7500 W Camp Wisdom Rd
Dallas, TX 75236
E-mail: http://www.sil.org/~simonsg
Kenneth S. Olson
SIL
7500 W Camp Wisdom Rd
Dallas, TX 75236
http://www.sil.org/~olsonk
Paul S. Frank
SIL
7500 W Camp Wisdom Rd
Dallas, TX 75236
E-mail: Paul_Frank@sil.org
[1]We wish to thank Fraser
Bennett, Kent Spielmann, and two anonymous referees for helpful comments on the
text, and Kazuaki Maeda, part of the TableTrans development team, for
programming assistance. An earlier version of this paper was presented at the
Fourth International Conference on African Languages, University of Maryland
Eastern Shore, November 4–7, 2004.
[2]Since the writing of
this paper, several additional documents have been published that also address
the question of best practices for digitizing audio recordings (Casey and Gordon
2007, IASA-TC03 2005, IASA-TC04 2004, Pohlmann 2006). The two reports from the
technical committee of the International Association of Sound and Audiovisual
Archives (IASA-TC03 2005, IASA-TC04 2004) recommend a minimum digital resolution
of 48,000 Hz, 24-bit for analog originals and note that 96,000 Hz, 24-bit has
become widely adopted in heritage/memory institutions. They recommend use of the
Broadcast Wave Format (BWF) file type, which is an extension of the WAV
format. |