Rehana Omardeen
Sampling signers in Providence Island
Who should be included in a representative sample of sign language users? Much sign language research has largely focused on native signers, deaf signers of deaf signing parents. However, as the breath of sign languages under study increases, it is becoming clear that focusing on deaf native signers may leave several blind spots. Among deaf signers, this group is a statistic minority, and in some settings native signers are near-impossible to find (Costello et al., 2008). Furthermore, deaf signers may only represent one part of the language community; in rural settings with high concentrations of genetic deafness, hearing signers can make up a key signing demographic (de Vos, 2016). I discuss the challenges of sampling signers of Providence Island Sign Language (PISL), a small rural sign language community concentrated around a small number of deaf people with highly heterogenous backgrounds, none of whom however are born to deaf parents. I critically examine in the context of PISL, who should be included in a representative sample and why.
Rural sign languages like PISL present a particular challenge for sampling. In one sub-type, village sign languages, genetic deafness may lead to a high proportion of deaf native signers available for sampling (Nyst, 2015). However, not all rural settings are the same. Rural sign languages can be used by small numbers of deaf signers dispersed over vast geographic areas (e.g. Schuit, 2010; Safar 2019). They can be used by just one deaf individual and their hearing network, or string together multiple connected networks of deaf signers (Reed, 2019; Hou, 2016). In these settings, vertical transmission from parent to child may be less important than horizontal transmission, and the role of hearing signers and conventionalized gesture may be particularly critical (Nyst, 2012; Reed, 2019; Mesh & Hou, in press). These rural sign language communities are so diverse that we must critically examine decisions of which signers to include in a representative sample.
In the Caribbean island of Providence, 13 deaf signers and their networks of hearing signers are dispersed across the island’s several villages. Signers range from 8 to 82 years old and all deaf signers have hearing parents. For most individuals, contact with other deaf people is infrequent. There is also is great variation in individual backgrounds, ranging from one multigenerational deaf extended family, to multiple individuals who have moved to the island and adapted their previous (home) sign systems to facilitate communication. There is a large degree of lexical variation among the deaf, however communication among signers is fluent.
I will present the challenges I faced in creating a corpus of spontaneous and semi-spontaneous conversation in PISL. Data collection took place as part of a documentation project over a 3-month period in early 2019. A total of 8 deaf and 4 hearing signers were recorded. I will evaluate the decisions made and discuss my successes and failures to create a representative sample with respect to (1) deaf signers and (2) hearing signers.
Firstly, PISL has relatively few deaf signers. This number shrinks further when taking into account minimum age for consent, as well as intelligibility and availability of the individual. I discuss exclusion criteria for sampling, given the diversity of backgrounds of the deaf in Providence. I will discuss issues such as the motivation behind including L2 deaf signers in my sample, as well as how I navigated interpersonal relationships among signers to maximize useful pairings in conversational settings. For example, pairing signers within the same family or within the same village resulted in a far richer, more natural conversational data than pairing signers who merely overlapped in age.
Secondly, I discuss the challenges of including hearing signers. While deaf-deaf interactions are often considered gold standard in sign language research, for most deaf people in Providence, deaf-hearing interactions are the norm. Interestingly, examining a single deaf signer in multiple settings can also reveal patterns; while one signer may use a seemingly idiosyncratic lexical variant, however this variant may be widely shared among his hearing network. Identifying suitable hearing signers proved challenging, especially as the communicative competence of hearing signers differed based on the type of interaction they routinely conduct in sign (e.g. sister versus employer).
Given that PISL have been described as heavily context dependent, a broad sampling of signers can give us a better idea of what kind of context counts when using the language. For example, previous literature has claimed that PISL signers do not use name signs and instead rely on vast shared knowledge to string together descriptions in personal reference (Washabaugh, 1986). In my sample, I have found that reference strategies differ depending on the interlocutor. Often signers use name signs when signing to (deaf or hearing) family members, yet they not use these name signs when talking to a deaf person from another village. Similarly, sampling widely allows us to capture the breadth of lexical variation in the language, as well as to examine how, by who and with whom, these variants are used.
Sampling broadly can benefit all language research, indeed mainstream linguistics research is evolving past the notion of an ideal speaker with more and more research on variation at the individual level (e.g. Yu & Zelou, 2019). In the case of rural sign languages, where there may be no “ideal signers”, this concern becomes more immediate, forcing us to revise ideas of what a representative sample looks like, and the breadth needed to faithfully capture the diversity of the language communities studied.