The global pandemic has highlighted experiences of isolation, upheaval, and uncertainty, while also revealing divisions within society. In this environment, it is crucial to examine biases that may exist both personally and within professional practices. This paper will focus on the importance of addressing assessment bias with children who are bilingual—defined here as having the need for two or more languages in order to successfully communicate in the home, school, and larger society (Kohnert et al., 2021). 

The Consequences of Assessment Bias 

The stakes are high when we get things wrong. One consequence of assessment bias is over-identification, which places an undue burden on children and families living with an inaccurate diagnosis and strains educational and health care systems whose resources may already be scarce.  

Assessment bias can also lead to under-identification—in which those children who do need additional support don’t receive it—thus, they may be more likely to fall farther behind. Under-identification can also stress educational and health care systems because delays in intervention can impact prognosis, treatment length, and outcomes. 

How To Reduce Assessment Bias 

We can never fully eliminate bias in assessment, given that power dynamics exist—and always will—in testing scenarios. However, we can reduce such bias in multiple ways. Two ways discussed here are (1) scrutinizing ourselves and our inherent biases and (2) scrutinizing the assessment tools that we choose to use. 

First and foremost, we need to engage in self-scrutiny to recognize biases that each of us brings into assessment. We as speech-language pathologists (SLPs) should consider how our own world lens is shaped by our prior experiences and influences the assessment approaches we select, who we involve in assessment, and our interpretation and reporting of results. Constant reflection and self-evaluation are essential for continued professional growth (Kohnert et al., 2023; see also the ASHA resources listed below).  

Another way to reduce bias is to scrutinize the tools that we use. Assessment tools that are produced in multiple languages often are translations or adaptations of tools that were originally produced in English. Whenever we use tools that were originally created for English-only speakers when we’re working with users of various additional languages, we automatically bring inherent bias into the assessment. Here are just two examples of such inherent bias: 

  • Direct translations of English grammatical tests may lack equivalent features across languages or may not best characterize the disorder in the target language. 
  • Translations of vocabulary tests that were standardized with English-only speakers may not increase from easy to difficult words as intended because word frequency and difficulty vary depending on the language.  

For the past 15+ years, my research team has been developing new assessment tools for monolingual and bilingual Vietnamese-speaking children in preschool and the early school years (for a review, see Pham, 2023).  

Here are some ways we’ve tried to reduce bias in our Vietnamese assessment tools. 

Identify language-specific properties

To help us build language assessment tools, we compiled a language corpus of more than one million words collected from Vietnamese children’s books and newspapers. We used the corpus to tell us about how frequently different speech sounds and speech sound combinations occur in Vietnamese, which allowed us to create nonword reading and nonword repetition tools tailored for the Vietnamese language. For assessment tools that required different levels of word difficulty, we confirmed low, mid, and high word frequency using this corpus. 

Consider the role of dialect

We considered phonological and lexical differences across the three main regional dialects of Vietnamese: northern, central, and southern. (Vietnamese dialects do not differ by grammar.)  

  • For sentence-level tasks like sentence repetition, we selected dialect-neutral words found across all dialects when possible or provided alternative word choices for northern and southern dialectal speakers.  
  • For tasks involving speech-sound manipulation—like nonword repetition—we used the subset of speech sounds and tones found across dialects.  

Approach assessment with historical and social awareness

To create tasks that could be used in Vietnam and with Vietnamese-speaking children in the United States, we avoided words that had a political undertone. This included words that entered daily use in Vietnam after the war ended in 1975 but that have not yet been adopted by diasporic communities. When engaging in discussions with community members who live in Vietnam or who live in the United States, we are also careful not to display symbols that might be offensive either to people in Vietnam (e.g., using images of the flag that the country used prior to the war) or to Vietnamese communities in the diaspora (e.g., using images of the current flag).  

Enable free and bilingual access

The cost of commercially available assessment tools can limit access, especially for users in countries with limited resources. Because we wanted our tools to be used by researchers and practitioners both in the United States and in Vietnam, we created a website (VietSLP, n.d.) where users can freely register to access our tools online. The website is bilingual with a button on each page to toggle between the two languages. For tasks that include audio stimuli, users can select the northern or southern dialect. Users can download record forms and administer tasks through our web-based interface. There are videos in English and Vietnamese that describe the research basis for the tools and how to administer and score each task. There is also a video for English-speaking SLPs on how to work with an interpreter to administer the tools. (See also the ASHA resource below on collaborating with interpreters.)  

How To Continue This Work: Improving Tool Accuracy and Scope  

Although we’ve spent a good amount of time and effort in developing assessment tools for Vietnamese speakers, we still have much work to do. We can do better in many ways—namely, by updating our tools as needed so we can further improve accuracy and, ultimately, reduce bias. Here are four ways to do just that: 

  1. Establish a bilingual reference point. When we started this work, there was little information on Vietnamese language acquisition. Thus, much of our tool-building focused on monolingual speakers of Vietnamese as an important first step. Although many of our tasks show diagnostic promise for monolingual children in Vietnam (Pham, 2023), we do not yet have diagnostic accuracy measures for bilingual children. 
  1. Avoid using cut-off scores. We do not recommend using cut-off scores derived for monolingual children in Vietnam with bilingual populations in the United States. For now, clinicians can use our tools with bilingual clients to monitor progress and/or compare results across children in their local settings. Our research team is actively collecting the large number of children needed to move the needle on diagnostic accuracy for bilingual populations. 
  1. Add contextualized measures. Our current assessment tools measure language at the speech-sound, word, and sentence levels. We recognize that they are decontextualized and that they are tasks typically used in academic settings. We need to add measures that capture language skills in naturalistic settings. Bilingual children in the United States use Vietnamese for specific purposes—often limited to home and family contexts. Measures of language development should better reflect the ways in which bilingual children need to use Vietnamese to communicate in their daily lives. 
  1. Go beyond identification. The tools currently available on the VietSLP website are meant for two possible purposes: (a) identify a language disorder and/or (b) measure progress and development. Although appropriate for assessment with clients, these tools are less useful when planning interventions with clients. They are meant to be used in combination with language sampling, observation, and parent/teacher report to co-develop goals and plans that improve a child’s overall communication. 

Current and Future Directions in Bilingual Assessment Research 

We at the Bilingual Development in Context Lab at San Diego State University are actively collecting data with bilingual children who speak Vietnamese and English in multiple states in the United States. In this effort, I’m working with colleagues at different universities to further reduce assessment bias. This work includes collaborating with Dr. Kerry Ebert at the University of Minnesota to develop nonlinguistic cognitive processing tools that tap into underlying skills of attention, memory, and processing speed.  

Another collaboration is with Dr. Elizabeth Peña at the University of California Irvine and Dr. Lisa Bedore at Temple University. Both projects are focused on building assessment tools in English created for and validated with bilingual children.  

In the future, I’d like to embark on building tools that better contextualize language development to (a) capture what bilingual families are already doing to support language development and (b) measure successful communication for specific purposes.  

Working as a collective, we can build better assessment tools to reduce bias and improve accuracy, which ultimately supports the well-being of children, families, and communities. 

ASHA Resources 

American Speech-Language-Hearing Association. (n.d.-a). Collaborating with interpreters, transliterators, and translators. https://www.asha.org/practice-portal/professional-issues/collaborating-with-interpreters/ 

American Speech-Language-Hearing Association. (n.d.-b).Cultural competence check-ins. https://www.asha.org/practice/multicultural/self/  

American Speech-Language-Hearing Association. (n.d.-c). Cultural responsiveness series. https://www.asha.org/practice/multicultural/cultural-responsiveness-series/ 

References 

Kohnert, K., Ebert, K., & Pham, G. (2021). Language disorders in bilingual children and adults (3rd ed.). Plural Publishing. 

Pham, G. (2023). A narrative approach to synthesizing research on Vietnamese bilingual and monolingual children. Journal of Speech, Language, and Hearing Research, 66(12), 4756–4770. https://pubs.asha.org/doi/10.1044/2023_JSLHR-23-00047  

VietSLP. (n.d.). Speech-language pathology resources for supporting Vietnamese children. https://vietslp.sdsu.edu/1