Hi everyone this is a recording of the 
Chirila Historical Linguistics Lab’s  
presentation at the recent Linguistic 
Society of America's Annual Meeting  
we're going to be talking about accessibility,  
discoverability, and functionality: an audit of 
and recommendations for digital language archives.
Thanks Claire. 

Okay, so we're going to start 
with some background about language archives  
so here's what we're going to talk about today: 
for our study we did a review of digital language  
archives, particularly for endangered and 
under-resourced languages. So as we all know a  
large portion of the world's languages are in some 
state of loss which makes language documentation  
and the accessibility of the materials that are 
being documented and preserved very crucial for  
future use and reclamation purposes. Digital 
language documentation is now the standard but  
they are very heterogeneous across archives,
and archives vary in standards and quality. 
So what is an archive? For our purposes we define 
an archive to be a repository of language data  
with the aim of preserving and disseminating 
those materials, so from Austin (2021):  
“an archive should appraise materials; preserve 
them long term; make their existence discoverable;  
and facilitate their appropriate distribution.”
 
Digital language archives have many pros and cons, 
and if you've ever used a digital language archive 
you're probably familiar with these. So a pro would  
be that they can be more compact than analog 
materials they allow for the digital manipulation  
of data — it allows for wider accessibility of 
data through using the internet and theoretically  
you can copy digital materials an infinite 
number of times without altering the original.  
However, some cons are that electronic storage 
mediums have poor longevity at times and they can  
still be vulnerable to environmental 
conditions and also more digital things like  
data corruption server issues and internet 
outages. Additionally the long-term readability  
is varying and as technology changes we need 
to make sure that digital language archives  
can can still be used even if technology is 
changing. 

So the problem we're looking at is 
digital language corpora are very heterogeneous. 
A lot of the standards of surrounding digital  
language archives were formalized a long 
time ago, and it was before much of the recent  
digital work was possible. Additionally, much of 
linguistic training focuses on producing  
material for archives, such as gathering language 
materials through fieldwork or something rather  
than working on materials from archives, where 
you're using archives as a starting point for  
some type of linguistic work. Additionally, some 
collections have fallen out of maintenance. You  
know websites are not actively maintained at 
times. 

So now we're going to move on to the review.
So to review the audit that we performed,
we surveyed roughly 50 archives from the  
perspective of an end user, rather than 
this perspective of a depositor who would  
deposit those materials in the archive. We focus 
primarily on three criteria, namely, (1) accessibility  
and findability, which includes what languages a 
user needs to know to access an archive and to  
navigate its collections; (2) discoverability 
which refers to the availability of the  
material within the collections and how users 
can navigate those archives and find, and finally  
(3) functionality which refers to how usable 
the materials are within those archives and  
what functions that the site and archive 
do or do not promote, that is, usability.

There are a few caveats to our survey, primarily 
archives do not receive the funding they need  
and a lot of this work is unpaid and very much 
a thankless job. We do not wish to discredit the  
value of this work — it's very important and it does 
need to be done. At the same time we do believe  
that a collection that is incomplete or imperfect 
is better than having no archive at all, so we  
do not wish to discourage archiving at all from 
the standards we set up for this recommendation.  
Finally, we need to clarify that our opinions 
and our positionality is from the endpoint of  
academics not necessarily the communities 
and language groups that might be using  
these materials for purposes of cultural 
reclamation or language education, so that  
needs to be acknowledged. 

At the same time, we 
do believe that our perspective offers some  
valuable insights into the nature of language 
archives, and points where they could improve.
We’re not naming archives directly, keeping 
this in mind we're not trying to name and  
shame individual archives these perspectives 
and these issues are happening across the board  
and pretty much everyone can be doing better.
So with this in mind we're focusing on issues  
that happen on a large scale. Furthermore, we 
really want to raise these points so that future  
archivists can raise the standards 
of what their work is going to be,  
and this in turn will help make sure 
the field is tenable for the future.

So for our results firstly we'll be 
focusing on accessibility results.  
Accounts and registration is the first tier 
of accessibility of any archive or any website.  
Some archives are open access, some require 
a free account, registration at most, and some  
require specific permission to access individual 
collections. Many archives combine these three  
modes of access. This is not an issue in that it 
often is used to respect the wishes of individual  
language communities and researchers but it can 
be clumsily executed in ways that unnecessarily  
impede the research of academics and language 
communities. For example, if for some archives  
account registration is built into Google Forms 
and meaning that lost and forgotten passwords  
can't be retrieved, so if you make an account and 
lose your password you have to make an entirely  
new one to access the archive. Furthermore, the time 
elapsed between account registration and actual  
access of the archive can be a matter of hours 
or sometimes days. Not all archives streamline the  
process to request permission to access individual 
collections in one archive they require that  
you mail in a physical copy of the request form 
which is not a viable option for every researcher  
and some archives were in some collections 
and those archives were only available  
through institutional access meaning you 
needed an institutional email or a password  
while this can provide access to academics we have 
to wonder what language communities who are not  
affiliated with academic institutions are losing 
out in terms of access to these cultural materials  
secondly interface language is a very 
important aspect of accessibility  
of the archives we suffer we surveyed only one 
third offered more than one language interface  
and it was particularly striking that of the 
available languages they tended to be in very well  
resourced European languages namely English French 
and Spanish plugging sites into Google Translate  
was not fully an option either because languages 
languages translated through Google Translate  
do not necessarily fully translate the 
entire site sometimes translations were  
inaccurate especially when languages were very 
different to european languages structurally  
and these translations often removed 
functionality meaning that some  
links became unclickable once 
they were plugged into filter
furthermore this emphasizes that the majority 
of languages and digital materials available  
to Indigenous users are in English even though 
English isn't the primary language for many and  
this further exacerbates existing divides in 
Indigenous communities where access to English  
language knowledge remains a critical issue 
we point to PARADISEC as a great example of an  
archive that makes its content available not only 
in multiple languages but in languages that are  
very relevant to a specific region for example 
Tok Pisin given that it focuses on Australian  
and Oceanic languages but other languages 
just did not have viable translations to  
language translation filters for example in this 
translation the name James Woodward is translated  
literally which makes it lose its connection to 
the tag in the archive so not only are we getting  
poor translations but we're losing functionality 
of the archive this issue is repeated again where  
we see that buttons and links no longer become 
are no longer clickable once they're translated  
in this cure gives translation example in some 
other instances only parts of the website were  
translated so we see the elicitation is translated 
into Kyrgyz once and then not in the other
disability accommodations are another critical 
aspect of accessibility for archives and indeed  
any other website it is essential that archives 
are designed in a way that is accessible to users  
who rely on assistive technology to navigate 
the internet because we are not experts or  
and we are not experienced in using screen 
readers to navigate the internet we felt  
that we were not able to properly survey archives 
for their compatibility with screen readers this  
wouldn't have been an honest reflection of our 
understanding of it but nevertheless we strongly  
recommend that principles of accessible web design 
which have long since been standardized should  
be followed most archives were friendly to users 
with color deficiencies but if you did suffer from  
low contrast between the font and background color 
which can be an issue for users who are colorblind  
or who have other visual impairments these 
issues are pretty easily rectified in terms of  
the overall scope of web design issues so 
we strongly recommend that they be followed  
overall our recommendations for accessibility 
include broadly streamline the process for account  
registration and permission requests but do not 
remove these impediments necessarily to implement  
a wider disarray of display languages when 
possible especially those that are relevant to  
the language communities hosted in these archives 
finally to follow principles of accessible web  
design these are baseline standards that will 
improve accessibility of archives across the board
our next category is discoverability one 
issue we saw with discoverability was in  
search functions and mislabeling users often 
can't see information about a collection before  
accessing it and opening it which makes browsing 
more cumbersome collections also frequently do  
not outwardly indicate their size or the types of 
files they store again you have to click into the  
into the collection to find out more 
about it also the ability to perform  
searches within collections was absent for 
most archives just searching for collections  
searches were often hampered by missing 
metadata incorrect tags and case sensitivity  
it allowed some collections to disappear in 
certain searches even if they were relevant  
also despite its importance for researchers 
only five archives allowed users to filter  
searches by media and file type which would 
hamper those who are looking for specific  
types of files and sometimes eaf files 
were mislabeled in browsers xml files
this is an example of an archive that did a good 
job of allowing for that sort of filtering and  
search this is the Alaska Native Language Archive 
and the picture shows that it allows the user to  
filter their search by whether a collection 
contains text files audio files or both
we also saw some problems in metadata how they 
were recorded there's little consistency between  
archives and how much or what kind of metadata 
is offered as well as how it's displayed  
even within archives collections vary 
wildly in the sort of metadata they provided  
and that makes navigation difficult because 
meta the metadata also affects the the searching  
for collections so can also allow certain 
collections to disappear in a targeted search  
metadata is also sometimes layered so there would 
be one set of metadata for a whole collection  
then a different set of metadata for 
individual folders within the collection  
and then sometimes even a whole different 
set of metadata just for individual files
we also saw some issues with site maintenance 
some sites required Yale VPN access for us  
researching through Yale some sites still require 
Adobe Flash Player which was discontinued as of  
December 2020 and blocked as of January 2021 
so accessing these sites is now impossible  
also server loss a movement can cause 
data loss especially if a website is not  
maintained properly and the 
Wayback Machine does not  
capture sufficient snapshots to fully restore 
lost archives and does not maintain links
many archives also suffered from pervasive broken 
links which makes navigation extremely difficult
okay so we offer these recommendations 
we recommend the archives offer  
more clear and detailed descriptions of 
collection content especially available  
before clicking into the collection we recommend 
that archives allow users to search by file type  
and they ensure the correct labeling of 
file type our next category is functionality  
in this category we mainly saw issues 
in site content structure and downloads  
most critically most archived that we looked at 34 
out of the 41 offered no option to bulk download  
collections only one file at a time this makes it 
extremely difficult and time consuming to download  
large collections like some that have over 15,000 
files and even smaller collections could be  
difficult to download in their entirety also file 
downloads cause a loss of nested file structure  
which again leads to more time spent trying to 
maintain format for large collections especially  
those that with imprecise file names this makes 
it especially hard to find matched content such as  
an audio track that has a separate 
transcription file and separate translation file
okay so for this category we recommend 
that archives offer bulk download  
options and also specifically offer the option 
shopping cart of downloads so that users can  
download specific files that they need all at 
once versus just downloading an entire collection
okay so moving on to our conclusions again 
to reiterate we as academics do not represent  
the entirety of all archive users and these 
results do not encompass the full breadth of  
what needs to be addressed especially for 
community members using language archives  
as starting points for language reclamation and 
pedagogy and we also reiterate that again archives  
often depend on volunteer labor and archives 
are often very underfunded and it usually takes  
a lot more funds and labor to have this type of 
long-term maintenance but we can still take steps  
to ensure that these language materials that 
often rely on archives to be preserved we  
can take these steps to make sure that they are 
well kept and safe and accessible for future use  
and so just to go over our overall recommendations 
again in terms of accessibility looking at  
accessible web design streamlining the process 
for accessing permission and registration  
and expanding interface languages of the archive 
sites and then in terms of discoverability  
having the ability to search by file type 
having more clearer and more consistent metadata  
and also making sure that file types are correctly 
labeled and then in terms of functionality  
having the option to bulk download or having that 
shopping cart download option will help streamline  
the process of using archive materials so going 
forward we want to look at i guess in terms of  
in linguistic circles there's the idea of corpus 
linguistics and that uses different corpora like  
COHA or COCA and we used corpora that were 
more describing under-resourced and endangered  
languages so a comparison between the corpora and 
materials between these types of corpora would be  
important and also we would like to look into 
what documentation archives have talking about  
their long-term backup plans so do they do they 
have some sort of system in case their website is  
taken down or in case links are broken and stuff 
because because this material is so important and  
because it's stored in these archives we 
can make sure that the materials are also  
backed up in a long-term and secure way and 
then we'd also like to look at mobile access of  
archives because in a lot of communities phones 
are the primary means of accessing the internet  
and so if these archives are available 
on the internet we need to make sure that  
tablet and phone access for these archives 
are also very usable and we would also lastly  
like to look at more experiences from 
end users especially community members  
who use language archives for language reclamation 
and pedagogy so thank you these are our references