What about broader impacts for child language acquisition research? A commentary on Kidd and Garcia (2022)

Introduction

To begin, I will clarify my vantage point for this commentary. I am a non-Indigenous linguist who works within the Chisasibi Child Language Acquisition Study (mun.ca/cclas). This project examines the L1 acquisition of Iiyiyiuyimuwin (Northern East Cree), an Algonquian language spoken within Eeyou Istchee territory in Northern Québec (see, for example, Henke, 2019; Henke & Brittain, 2022). In this response, I draw from my own experiences to address issues pertaining most specifically to contexts involving Indigenous languages in Canada and the United States – henceforth, North American (NA) languages – not necessarily throughout the world. Approximately 300 NA languages were spoken prior to European contact (Mithun, 1999), and such languages occupy a special position in the realm of language acquisition research. Many have linguistic properties that differ dramatically from those in more-studied languages, and science has uncovered relatively little about how NA languages are acquired. Probably all NA language communities are experiencing language loss on some level, primarily due to the legacy of colonization, and few are still acquired by children as a first language in a traditional manner.

Kidd and Garcia (2022) have cogently articulated some problems for science associated with the lack of language diversity in L1 acquisition research. They make a valuable contribution by providing (more) empirical evidence demonstrating that linguistic science is lagging in its mission to understand the human capacity for language. Their article moves the conversation further in a much-needed direction, but I want to highlight more rarely considered issues.

Much of the work surveyed by Kidd and Garcia comes out of the United States, where the National Science Foundation (2021) uses the following two criteria in evaluating research proposals: intellectual merit and broader impacts. The former concerns areas such as contributions to scientific theory. The latter entails fuzzier – but, in my opinion, more important – implications such as contributing to the social and cultural needs of particular communities. These criteria shape much of today’s L1 acquisition research landscape, yet Kidd and Garcia focus exclusively on stakes related to intellectual merit. Crucially, the lack of language diversity also creates problems for broader impacts, especially regarding NA language communities.

Science knows even less than we think

First, an additional complication related to intellectual merit. Kidd and Garcia employ a binary operationalization of language coverage: There is published research on Language X or there is not. This sensible methodology generates valuable insights, but it also obscures the depth of the diversity problem.

Dr. Melvatha R. Chee and I (to appear) have attempted to survey all published academic studies pertaining to the L1 acquisition of NA languages. We identified approximately 90 studies involving about 30 total languages or varieties. Many are not published in major acquisition journals, as Kidd and Garcia suspect (2022, pp. 18–19). The numbers we found ostensibly boost the state of diversity in language coverage, but there is a trade-off involved: Many of the findings are not grounded in contemporary standards for scientific research, and they often entail a host of foundational issues related to sampling and analysis – not to mention reproducibility and replicability.

For instance, from the sizable Uto–Aztecan language family, Chee and I found just two published studies with information related to child or child-directed speech – one for Comanche (Casagrande, 1948) and another for Hopi (Titiev, 1946). A two-and-a-half-page brief, the Hopi piece contains just one sentence commenting on child vocabulary and two on the omission of inflectional elements in child speech. The study also furnishes an 11-item list contrasting word forms in child and adult speech. Along with this dearth of information, the author provides a single vague hint about its source: ‘‘This phenomenon [of child-specific vocabulary] has not been previously reported, but the writer was informed that Hopi infants learn to use many words that are not employed by adults’’ (Titiev, 1946, p. 89).

To be clear, I believe such findings are nonetheless valuable. In an impoverished research landscape, every bit of information has potential intellectual merit. But in appraising and increasing the diversity of language coverage, science must contend with issues related to quantity and quality.

The lack of diversity impacts applications for L1 development

Despite rapid and widespread language loss, children in some NA language communities still acquire their traditional language as an L1. In some communities, children may speak a variety that blends the language with another such as English or French. In either case, a common need emerges. Researchers, clinicians, and families have called for linguistically and culturally appropriate methods for assessing and supporting the language development of children (e.g. Peltier, 2011). Here, the lack of language coverage in L1 acquisition research circumscribes the field’s broader impacts: Because we do not know enough about how children acquire NA languages, science has a difficult time helping to produce what communities say they need.

Consider an anonymized example drawn from my experience, which illustrates problems some communities face. In one community, a few thousand people across all age groups still speak traditional Language Y, and even more speak a variety mixing Y with English. Speech-language pathologists (SLPs) tend to come from outside the community and speak neither. These clinicians say they need methods and tools grounded in empirical data so they can better serve the community. For instance, they want to know the typical path of development for both language varieties, including important milestones, so they can more effectively identify potential language delays and disorders. They want faster, more efficient tools, such as word- and sentence-repetition tasks, that would enable them to serve more children and families. These SLPs are few in number and lack resources that are tailored linguistically and culturally for the community, so they do their best to adapt those made for English. Right now, L1 acquisition science simply does not have the data to support the production of the resources that these clinicians and local families need. Although a relatively small body of published academic work has developed around similar issues (e.g. Allen et al., 2019), the field has a long way to go in making broader impacts through diversified language coverage.

The lack of diversity shortchanges language revitalization efforts

Many NA language communities are fighting language loss through a range of revitalization programs. One particular collection of efforts intersects directly with L1 acquisition research: language-immersion programs for young children, such as language nests (see, for example, Hinton & Hale, 2001). As another anonymized example, consider community Z, which has had no traditional L1 speakers for about two generations. Language Z is radically different from English, as for instance, a given verb root has scores of polymorphemic inflectional forms. Through intensive efforts, the community has educated a small group of fluent L2 speakers, who now work in the language-immersion daycare. Here again, the lack of language coverage in acquisition research impairs broader impacts.

For starters, there is the matter of language input, which is exponentially more important in language revitalization situations (O’Grady et al., 2021). Immersion programs need to produce language-intensive environments for hours each day, providing rich input for children in contexts where they might receive no additional exposure to the language. In the case of Language Z, no scientific data exist about the input children traditionally received. The daycare teachers did not grow up in the language and themselves range in proficiency, without complete command of all possible inflectional forms. The fundamental problem is that teachers want to know how they should talk to the children in order to successfully engender a new generation of speakers. They ask many questions about what to do, including: How can they know if they are giving children the right input? How morphologically simple or complex should the input be? Which verbs should they use, and in which inflectional forms?

Another issue arises with assessing child language development in the daycare. To date, scientific literature has offered no language-specific insights into how Language Z is acquired, and the community lacks information about the path of acquisition for children in traditional circumstances – let alone in a language revitalization context. Once again, teachers and language planners have a variety of questions. What would typical development look like in Language Z? What kinds of milestones should children be reaching and when? How can they assess child speech to determine whether the program is achieving its goals and make necessary adjustments to the program?

Conclusion

I have only scratched the surface here, and particularly in language revitalization contexts, the stakes can be incredibly high while resources may be low. Teachers, language planners, administrators, clinicians, parents, and others are doing their best and looking for whatever support they can find. The types of questions mentioned here have few readily available answers from the field of L1 acquisition, in large part because of shortcomings in language coverage. In our effort to appraise and rectify this diversity problem, we must also consider the stakes for broader impacts – not just those for intellectual merit.