This week on the ELAR blog, ELDP grantee Richard T. Griscom gives us some insights on his experiences conducting fieldwork remotely – something he also recently discussed on ELAR depositor Martha Tsutsui Billins’ podcast Fieldnotes. Richard’s ELAR collection ‘Documentation of Isimjeeg Datooga’ can be accessed here.
By Richard T. Griscom
The COVID-19 pandemic has brought a lot of uncertainty into our lives. For those working with endangered language communities the travel bans, quarantine requirements, and other restrictions that come with the pandemic may prevent us from engaging in many of our regular language documentation and description activities. During a recent gathering of Linguistics in the Pub, panel members discussed various ways that linguists in an academic setting can continue work with language communities at a distance during the pandemic, and a number of posts on social media confirm an increased interest in “remote fieldwork” methods. Today I will discuss how to work remotely with a member of a speech community to elicit linguistic material, based on my experiences working with the Hadza and Asimjeeg Datooga communities of Tanzania during projects funded by the Endangered Languages Documentation Programme.
The phrase “remote fieldwork” may have an unpalatable ring to it for some, especially given recent calls to decolonize academic research and institutions of higher education. The phrase may evoke an image in your mind of a disconnected and disinterested researcher extracting information from a speech community, but in reality any remote workflow can be used as an opportunity to build community capacity and promote greater community agency in the language documentation process, which provides benefits both for research and the community (Czaykowska-Higgins 2009). In the context of language documentation, what “remote” often means is that someone within the community is completing tasks in cooperation with a physically distant specialist, who may be a member of the community or an outsider. As we shall see later, such an arrangement often involves an exchange of knowledge and skills.
First, a couple of important caveats. It may seem obvious, but you can’t easily begin remote work with a community that you don’t already have a good working relationship with. For some communities, a strong relationship with an outsider can take months or even years of in-person interaction to develop. If that relationship doesn’t already exist, then don’t expect to be able to create it remotely! Secondly, prior to engaging community members in remote work, you should carefully evaluate the anticipated impact of your work on the speech community during this challenging and potentially dangerous time, and consider project tasks such as archiving or data augmentation which can enhance the resources you have already created with minimal impact to the community. Keeping these two caveats in mind, let’s now turn to remote elicitation.
Elicitation is not typically associated with language documentation, but it often constitutes at least a subset of language archive material (Lüpke 2010; Woodbury 2003), and for those researching particular linguistic phenomena it can be a useful method for quickly collecting a targeted range of data. Traditionally elicited linguistic data were recorded with analogue media such as cassette or reel-to-reel tapes, and transcriptions were handwritten in a notebook (Lane and Carylye 2013: 202; Bowern 2008: 20). Today it is increasingly common to create both in digital format (Thieberger and Berez 2011; Margetts and Margetts 2011; Griscom and Otero 2019). Here I will focus on collecting digital recordings of elicited data.
How do you facilitate the creation of a remotely recorded elicitation session? As mentioned earlier, if you are not there in person then someone else must make the recording. That someone can be the speaker recording themselves or another person. For the sake of simplicity, let’s call that person the data collector. The one who is being recorded and is producing the elicited items will be referred to as the speaker. If the data collector and the speaker are the same person, then that simply means that they are recording themselves. This guide is written for the person who is acting as the facilitator, or the one who supports the creation of the recording remotely. As part of your role as facilitator, you must help the data collector learn how to make a good recording with the equipment that they have access to and how to conduct a simple elicitation session. If you already have a good relationship with a community member who knows these things, then you are in luck! You can likely conduct remote elicitation without providing any additional training. If the data collector does not have much experience with elicitation, then set aside some time specifically to train them before attempting to collect any data.
When training a data collector, try to cover all of the details related to conducting a good elicitation session. Teach them how to use their recording device, how to position the microphone, how to select a good recording environment with little background noise, and how to take notes. Practice producing words with repetitions and container phrases. You can even perform a mock elicitation session, during which you give the data collector the opportunity to practice reacting to a speaker who produces an unexpected form or produces forms in an alternate order. Also, make sure to train the data collector to collect metadata for the speaker and the recording session according to the metadata standard that you have established for your project. Metadata is crucial for archiving and for any research that involves extra-linguistic factors (Good 2002; Kendall 2011). Finally, use the training as an opportunity to support the data collector in pursuing their own goals, both within and outside the scope of language documentation, and consider encouraging the data collector to create their own recordings with the community.
It is also important to make sure that your data collector and speaker are being compensated for their contributions according to the standards of your project. If you normally provide monetary compensation, then you need to work with the data collector to find a way to send money so that they can distribute it to the speakers who are being recorded. In many regions of the world, it is now possible to send money through online services such as Transferwise, WorldRemit, Remitly, and Xoom. If the data collector has a bank account, then a bank transfer should be possible. It is also often possible to send money to a mobile banking account, which the data collector may be able to use to withdraw cash in areas where there is no bank.
A typology of elicitation sessions
Broadly speaking, there are two types of remote elicitation methods: synchronous remote elicitation, and asynchronous remote elicitation. The former resembles a traditional elicitation session in that the facilitator and the speaker (or data collector) communicate with each other using VoIP or web-conferencing software, and a recording of at least some portion of that interaction is made by the data collector. This type is labeled “synchronous” because the interaction between the facilitator and the speaker is simultaneous with the recording of the speaker. Asynchronous elicitation, on the other hand, often consists of two stages: first the facilitator communicates to the data collector what should be elicited from the speaker, either through a message or live communication, and then the data collector creates a post-hoc recording of the speaker.
Synchronous elicitation is only possible if both you and the data collector have access to a fast and stable internet connection and either a computer or mobile device. In addition, it is strongly recommended that an independent device be used for recording. The reason for this is that the audio of your remote communication will not be of very high quality, and the recording that you produce may not be useful or meet the standards for the language archive or repository in which you would like to preserve your data. The ideal independent recording device is an audio recorder, such as a Zoom H5 or Tascam DR-100, used together with a headset microphone such as Shure SM35. A mobile device with an internal microphone, such as an Android or iPhone smartphone, can be a good substitute when no audio recorder is available. Two Android apps for creating audio recordings are RecForge Pro, and Easy Voice Recorder, and two comparable iOS apps are AudioShare and Awesome Voice Recorder.
The software that you use to communicate with the data collector or speaker will be a form of either Voice over Internet Protocol (VoIP) or web-conferencing software technology, such as WhatsApp, Skype, Facetime, Zoom, Google Meet, or Microsoft Teams. Which particular software you use will depend on what is easy and accessible for the data collector. To create a digital space for the data collector to store recordings, you can prepare an account with a cloud storage service, such as Dropbox, Google Drive, OneDrive, or iCloud. Make sure to train the data collector how to upload data to that particular service and practice together.
After you have prepared all of the software, trained the data collector, and identified a speaker to work with, there is still one more important step that you must consider prior to recording anything. Has consent to record been obtained from the speaker? Consent should be obtained following the standard process that you use for in-person recordings in your project, in order to maintain consistency, and the data collector may need to assist in explaining aspects of consent to the speaker in person. Video or audio consent may be a better method than written consent. Make sure to practice these methods with the data collector if they do not already have experience obtaining consent.
Asynchronous elicitation is possible if both you and the data collector have access to some kind of internet connection, even if it is intermittent and unstable, and either a computer or mobile device. The quality of the internet connection will determine the best method for communicating with the data collector.
One primary difference between asynchronous and synchronous elicitation is the function of the communication: rather than conducting the entire elicitation session live, the facilitator must communicate with the data collector about the specifics of the session prior to recording. These specifics can include the speaker to be recorded (e.g. a known individual, or an individual of a certain age, gender, or background), the use of repetitions and container phrases, and the data to be recorded. The facilitator most likely will need to send the full list of items to be recorded in a text format that can be used by the data collector, such as a PDF or spreadsheet file.
A second difference between asynchronous and synchronous elicitation is that an internet connection is not required at the time of recording. This means that the communication between the facilitator and data collector can happen at a different time and place than the recording, and the recording can be created in a remote area with no internet connection or electricity at all.
For those working with data collectors who don’t have access to an internet connection that is stable enough to support VoIP or web-conferencing, asynchronous elicitation can still be achieved. The speaker or data collector only needs access to an internet connection in order to receive text instructions and send recordings, and such a connection does not need to be as stable or as fast as the connection required for VoIP or web-conferencing. It may be helpful to use a cloud service such as DropBox that supports the resumption of data transfers even after the internet has been disconnected. This feature allows the data collector to load large files even with a slow or unstable connection.
There is one generalization that cuts across both remote elicitation methods: as the quality of the internet connectivity decreases and the types of communication become more restricted, the amount of prior training and experience the data collector must have for successful elicitation increases. Linguistic elicitation is quite an unnatural activity. If the speaker you are working with has no experience with linguistic elicitation, then either you or the data collector will need to provide a detailed explanation of some kind and guide them through the process. Communicating such details through a web-conferencing session with video may be straightforward, but doing the same through text messages will likely be quite difficult. This means that you will need to provide the data collector with sufficient training prior to the elicitation session so that they can effectively guide the speaker through the process.
Remote elicitation is predicated on a pre-established relationship between a specialist and a community member, and it requires specialized training before elicitation can begin. An internet connection and computer or mobile device are also required for communication with your collaborator(s). If you are able to meet these conditions, and you have carefully considered the impact of your work on the community during the pandemic, then remote elicitation may provide a good opportunity to advance towards your own language documentation goals while also transferring skills and knowledge to the community and offering a starting point for community-led documentation.
Bowern, Claire. 2008. Linguistic Fieldwork: A Practical Guide. Palgrave Macmillan.
Czaykowska-Higgins, Ewa. 2009. Research Models, Community Engagement, and Linguistic Fieldwork: Reflections on Working within Canadian Indigenous Communities. Language Documentation and Conservation3(1). 15–50.
Friederike Lüpke 2010. Research methods in language documentation. In Peter K. Austin (ed.) Language Documentation and Description, Vol 7, 55-104. London: SOAS.
Good, Jeff. 2002. A Gentle Introduction to Metadata.Available at http://www.language-archives.org/documents/gentle-intro.html. Accessed June 22nd 2020.
Griscom, Richard T. and Otero, Manuel. 2019, February 29th. The digital notebook: a method for the rapid processing of elicited linguistic data. Presented at the 6th International Conference on Language Documentation and Conservation (ICLDC), Honolulu.
Kendall, Tyler. 2011. Corpora from a sociolinguistic perspective. Revista Brasileira de Linguística Aplicada 11(2). 361–389. doi:10.1590/S1984-63982011000200005.
Lane, Cathy and Carlyle, Angus (eds.). 2013. In the Field: The Art of Field Recording. London: Uniformbooks
Margetts, Anna & Andrew Margetts. 2012. Audio and video recording techniques for linguistic research. In The Oxford Handbook of Linguistic Fieldwork. Oxford University Press.
Thieberger, Nicholas & Andrea L. Berez. 2012. Linguistic Data Management. In The Oxford Handbook of Linguistic Fieldwork. Oxford University Press.
Woodbury, Anthony C. 2003. Defining Documentary Linguistics. Language Documentation and Description, vol. 1. London: Hans Rausing Endangered Languages Project.