A 10-Week Systematic Comparison of Three Aided Language Applications for Supporting Vietnamese-English Vocabulary Development: A Descriptive Case Study

Background

AAC research has largely centered on monolingual English speakers, and many aided language applications default to English syntax, vocabulary, and language organization. Recent work is clarifying how aided AAC can support bilingual communication, including code-switching and translanguaging between a home language and English. Code-switching alternates languages within a message to increase efficiency or meaning, whereas translanguaging reflects an integrated bilingual repertoire without strict language boundaries (García & Wei, 2014).

Vietnamese-speaking families face additional barriers, including limited Vietnamese vocabulary, difficulty representing tonal features in English-dominant voice settings, and limited clinician preparation to implement bilingual AAC and provide consistent bilingual aided language input (ALI). Although bilingual children with autism spectrum disorder (ASD) can acquire two languages without adverse effects (Hambly & Fombonne, 2012; Kay-Raining Bird et al., 2016), AAC applications may still constrain bilingual development due to system limitations and limited partner training (Soto & Yu, 2014; Soto et al., 2026).

Purpose

This 10-week single-case study examined an autistic child’s ability to combine core and fringe vocabulary across Vietnamese and English, measured by the Total Number of Words (TNW) and Number of Different Words (NDW) communicated multimodally. The study also used a five-point ease-of-use rating scale (EOU) to compare three AAC applications (Flexspeak, Proloquo4Text, Avaz).

Methods

A descriptive case study was conducted over 10 weeks (three 60-minute sessions per week) in a university clinic and at home. At study onset, the child’s highest and primary level on the Communication Matrix was Level VI-Abstract Symbols. The child arrived with school-provided TouchChat LOC 60 and consistently produced spontaneous gestalts of 1–3 word utterances for contextually confined purposes.

In the university clinic, communication spaces were engineered to accommodate sensory needs, with activities to stimulate expressive language and maintain interest. The parent was taught how to provide consistent bilingual ALI with each application.

Data included usability ratings, session notes, and audiovisual recordings; 30 sessions were recorded and reviewed. Quantitative measures included TNW and NDW in Vietnamese and English and the EOU reflecting independent access across home and clinic (1 = not accessible to 5 = fluent/automatic). The EOU was completed by three raters (mother, behavior technician, clinician) separately, and the clinical supervisor reviewed selected sessions. Qualitative documentation captured vocabulary exploration, translanguaging behaviors, and communication partner strategies. Data were analyzed using the EOU rating scale for each application.

Results

The child completed nine sessions per application (Avaz, Proloquo4Text, Flexspeak). TouchChat LOC 60 was available in all sessions and offered no Vietnamese options.

Avaz produced the highest output and lexical diversity (TNW = 89; NDW = 51). Proloquo4Text also yielded high output (TNW = 83; NDW = 42), but a substantial portion reflected non-communicative typing. Flexspeak showed more modest gains (TNW = 24; NDW = 14). TouchChat usage was also documented, but output was limited (TNW = 12; NDW = 7).

Vietnamese production differed across systems: TouchChat and Flexspeak produced no spontaneous Vietnamese words (TNW = 0), Proloquo4Text produced minimal Vietnamese (TNW = 2), and Avaz supported the most Vietnamese (TNW = 33).

Qualitatively, Avaz supported the most consistent bilingual communication, with more spontaneous exploration and intentional messages. Proloquo4Text enabled substantial text-based output but focused on restricted-interest content. Flexspeak supported emerging bilingual communication with moderate engagement and variable independence.

Results on the EOU differed: Avaz (M = 3.8) and Flexspeak (M = 3.7) were highest, TouchChat was moderate (M = 3.0), and Proloquo4Text was lowest (M = 1.5), indicating greater navigation demands and increased need for partner support. Krippendorff’s alpha was used to calculate interrater reliability between three raters and there was perfect agreement (α = 1.00).

Discussion, Limitations, and Clinical Implications

AAC system design influenced expressive output, multilingual access, and usability. Avaz showed the strongest support for bilingual communication and Vietnamese access, likely due to easier editing navigation and stronger multilingual vocabulary availability. Proloquo4Text showed promise for generative text-based communication but was less easy to use for this beginning communicator and had more restricted-interest output. Flexspeak offered relatively strong usability with smaller language gains, while TouchChat had limited expressive output and no Vietnamese access.

Limitations include the single-case design, limited generalizability, and the inability to draw causal conclusions. Additionally, exposure to AAC outside sessions and subjective interpretation of qualitative data may have influenced results. Future research should include larger samples and more controlled comparative designs.

Clinically, results support feature matching that prioritizes access to both languages, efficient partner bilingual ALI, and flexible language switching. Translanguaging-informed intervention, support for user autonomy, and ongoing system optimization with caregiver input may improve bilingual communication.